A Guide to Networking Macs
!
James K. Anders
A ^ Easy-to-Use
Mac Symbol Library
for Simplified
Network Design
Foreword By Guy Kawasaki
Live Wired
A Guide to Networking Macs
y ^ '*L
James K. Anders
j)
I'j.
n 11/97
^BOOI^
$ 2.00
sofZA^r
nAYDEN
Live Wired: A Guide to Networking Macs
Copyright © 1993 James K. Anders
All rights reserved. Printed in the United States of America. No part of this
book may be used or reproduced in any form or by any means, or stored in
a database or retrieval system, without prior written permission of the
publisher except in the case of brief quotations embodied in critical
articles and reviews. Making copies of any part of this book for any
purpose other than your own personal use is a violation of United States
copyright laws. For information, address Hayden, 1 171 1 N. College Ave„
Carmel, IN 46032.
Librar>^ of Congress Catalog No.: 93-77320
ISBN: 1-56830-015-8
This book is sold as is, without warranty of any kind, either express or
implied. While every precaution has been taken in the preparation of this
book, the publisher and authors assume no responsibility for errors or
omissions. Neither is any liability assumed for damages resulting from the
use of the information or instructions contained herein. It is further stated
that the publisher and authors are not responsible for any damage or loss
to your data or your equipment that results directly or indirectly from your
use of this book.
95 94 93 4 3 2
Interpretation of the printing code: the rightmost double-digit number is
the year of the book’s printing; the rightmost single- digit number the
number of tlie book’s printing. For example, a printing code of 93-1 shows
that the first printing of the book occurred in 1993.
Trademark Acknowledgments: All products mentioned in this book are either
uademarks of the companies referenced in diis book, registered trademarks of the
companies referenced in this book, or neither. We strongly advise that you inves-
tigate a particular product’s name thoroughly before you use the name as your own.
O
Dedication
To Ada
Time passes...
but memories remain.
Credits
Development Editor
Karen Whitehouse
Production Editor
David Ciskowski
Technical Reviewer
Richard Leach
Cover Designer
Kathy Hanley
Designer
Scott Cook
Indexer
Jeanne Clark
Production Team
Diana Bigham, Tim Cox, Mark Enochs, Tom Loveman,
Michael J. Nolan, Joe Ramon, Carrie Roth,
Mary Beth Wakefield, Barbara Webster
Composed in Utopia
0
About the Author
James K. Anders
Jim Anders is a Senior Consulting Engineer with Computer Meth-
ods Corporation, where, for the last five years, he specialized in
desktop systems integration, focusing on Macintosh/VAX integra-
tion and CAD/CAM technologies. Through Computer Methods,
Mr. Anders teaches the seminar “Integrating Apple’s Macintosh
into VAX System Networks,” for Digital Equipment Corporation.
He has provided consulting services for companies such as Apple
Computer, Inc., Boeing Aerospace, Colgate Palmolive, Digital
Equipment Corporation, Du Pont, Mobil Oil, and Union Carbide
Corporation.
Jim Anders writes and lectures on Macintosh networking, systems
integration, and CAD/CAM topics in a variety of forums, such as
the Mactivity networking conference and the Macintosh Summit
Conference. He is also the author of Technical Drawing with Claris
CAD and The Macintosh CAD/CAM Book. He also writes reviews
and articles for MacUser magazine.
His background includes over fifteen years experience with the
implementation and administration of networked, integrated
CAD/CAM systems. He was an engineering staff member of
RCA Missile and Surface Radar, where in 1986 he spearheaded a
successful effort to integrate Apple’s Macintosh with their
multivendor network. Previously, Mr. Anders worked for RCA
Laboratories in Princeton, New Jersey, and for various companies
in the medical and semiconductor manufacturing industries.
We Want to Hear
from You
What our readers think of Hayden is crucial to our sense of well-
being. If you have any comments, no matter how great or how
small, we’d appreciate your taking the time to send us a note.
We can be reached at the foUowing address:
Hayden
11711 N. College Ave.
Carmel, IN (yes, Indiana — not California. . .) 46032
(317) 573-6880 voice
(317) 571-3484 fax
If this book has changed your life, please write and describe the
euphoria you’ve experienced. Do you have a cool book idea? Please
contact us at the above address with any proposals.
o
Acknowledgments
Books are collaborative efforts. Even though there’s one name on
the cover, there are many people who deserve credit.
Guy Kawasaki deserves a special note of thanks. It was his contin-
ued insistence and cajoling that prompted me to expand and
formalize a set of lecture notes that were used to help explain
networking concepts to attendees of the Macintosh Summit
Conference, held at the University of California at Santa Barbara.
Joe Dallatore provided valuable assistance with the Management
and Troubleshooting chapter and with the technical aspects of the
manuscript.
Sunny Anders spent countless hours reading and re-reading every
page of this book — making sure that everything made sense.
Gloria Anders created the wonderful cartoons used throughout the
book.
Thanks to Bill Sturdevant, and Silk City Software, for their timely
and professional development of the PICTviewer application
included on the disk that comes with this book.
I also would like to acknowledge the efforts of Garry Hornbuckle of
Apple Computer, and the many representatives from the various
Macintosh and AppleTalk networking companies.
Thanks also to Stephen and Maria Hull.
A special thanks to Karen Whitehouse and David Ciskowski of
Hayden Books for their tireless efforts providing the necessary
direction and editing to take this project from a big Microsoft Word
file and a bunch of MacDraw Pro documents into the finished
product that you see before you.
Finally, thanks to A1 Cini (of Computer Methods Corporation), who
nurtured and encouraged many of the ideas and concepts outlined
in this book. I would like to offer my undying gratitude for those
individuals — Frank Monzo, Anthony Caraffa, Ruth Logue and
Vickie Briggs — at CMC who make my job so much easier.
Jim Anders
Feasterville, Pennsylvania
March, 1993
Foreword
For a computer that was supposed to be for “the rest of us,”
Macintosh has progressed (I use the term loosely) until it's now a
computer for “the best of us.” 1 often reminisce about the days
when LocalTalk was AppleTalk, and AppleTalk was nothing more
than a long printer cable to reduce the cost of a LaserWriter: “You
see, if we buy a $7,000 printer, seven people can share it on an
AppleTalk network, so it really only costs $1,000.”
That was then — and this is now. IBM and Apple are allies. There's a
twelve month warranty for Macs in the United States. The Berlin
Wall is gone. “Macintosh Portable” isn't an oxymoron anymore.
And Macs are connected on networks in organizations ail around
the world — even on half breed, mulatto, hapa-haole networks of
Macintoshes, IBMs, DECs, HPs, and other superfluous mini- and
mainframe computers.
In this confusing world, all 1 can say is, “Thank God for Jim
Anders.” I’ve sat through many a presentation about networking
and communications and Jim is the only person I can understand.
He's so good at it that I usually beg him to explain networking and
communications to a Macintosh conference I run for desktop
publishing weenies.
Indeed, part of the reason this book exists is because I pounded on
Jim to write a book that would explain networking and communi-
cations to the rest of us. I don’t know how he positions Live Wired,
but in my mind, it’s the The Mac is Not a Typewriter (by Robin
Williams) of networking and communications books. If you know
that book, you’ll know how good Live Wired is. If you don't,
trust me.
Guy Kawasaki
Macintosh Curmudgeon
0
Table of Contents
at a Glance
Foreword ix
Introduction 1
I Networking Fundamentals
1 How Does Communication Take Place? 9
2 How Does Networking Take Place? 17
3 Network Diagramming with NetPICTs 31
4 Networking Concepts 47
5 Common Network Components 65
II Macintosh Networking
6 Macintosh Services 99
7 Macintosh Formats 115
8 Macintosh Transport: AppleTalk 127
9 Macintosh Media/Cabling 179
III Multivendor Networks
10 Living in an Intel /DOS World 211
11 UNIX Connectivity 231
12 DEC VAX Connectivity 251
13 IBM Connectivity 269
IV Network Design, Implementation,
and Management
14 Design and Implementation 283
15 Management and Troubleshooting 313
16 The Future of Macintosh Networking 337
Glossary of Networking Terminology 341
Appendix A: NetPICT Encyclopedia 369
Appendix B: Listing of Macintosh
Networking Companies 383
Index 395
How to Use the Disk 433
Table of Contents
Acknowledgments vii
Foreword ix
Introduction 1
I Networking Fundamentals
1 How Does Communication Take Place? 9
Idea 10
Expression 10
Transport 11
Medium 11
Matching Layers 13
Conclusion 16
2 How Does Networking Take Place? 17
Idea > Services and Programs 18
Expression > Formats 19
Transport > Protocols 22
Medium > Cabling 25
Conclusion 30
3 Network Diagramming with NetPICTs 31
Introducing the NetPlCT symbol 31
Swapping Layers 36
Matching Layers 37
Clients and Servers 39
The OSl Reference Model and NetPICTs 40
A Seven Layer OSI Example 44
Conclusion 45
o
Live Wired
4 Networking Concepts 47
Types of Networks 47
Network Frames 56
Network Signaling 57
Conclusion 63
5 Common Network Components 65
LANs & WANs 65
Repeaters 66
Bridges 69
Routers & Brouters 73
Gateways 86
Miscellaneous: Hubs, Concentrators, and So Forth 90
Network Topologies 91
Conclusion 96
II Macintosh Networking
6 Macintosh Services 99
AppleShare (AFP) 99
System 7 File Sharing 101
Other AFP Clients and Servers 102
Print Services (PAP) 103
Apple Open Collaboration Environment (AOCE) 105
Terminal Services 106
Data Access Language (DAL)
and Other Database Services 108
Mail Services 110
Conclusion 113
7 Macintosh Formats 115
ASCII Text and Word Processors 116
MacPaint and PICT Formats 118
PostScript 119
TIFF and GIF 120
Binary 121
Document Interchange 122
Conversions 124
Conclusion 126
Live Wired
8 Macintosh Transport: AppleTalk 1 27
The AppleTalk Protocol Family: An Overview 127
How Does AppleTalk Work? 137
AppleTalk Phase 1 and 2 148
How Does the Chooser Work? 155
AppleTalk Routing 166
Conclusion 178
9 Macintosh Media/Cabling 1 79
LocalTalk/Phone-Type Connectors 179
Ethernet/EtherTalk 187
Other Cabling Systems 196
WAN Media 203
Conclusion 207
III Multivendor Networks
10 Living in an Intel/DOS World 211
Services: Application -Based 211
Formats; ASCII, EPS, Binary Compatible 213
Transports: AppleTalk (PhoneNET PC),
MacIPX, DECnet, TCP/IP 218
Cabling Options 227
Conclusion 230
11 UNIX Connectivity 231
Services: FTP, NFS, AFP, X- Windows, TELNET 231
Formats 242
Transports: AppleTalk, TCP/IP 244
Cabling: Ethernet, LocalTalk, Token Ring 248
Conclusion 249
12 DEC VAX Connectivity 251
Services: PATHWORKS for Macintosh 251
Formats 257
Transports: AppleTalk for VMS, DECnet, LAT 260
Cabling: LocalTalk and Ethernet 266
Conclusion 268
Live Wired
13 IBM Connectivity 269
Services; Mainframe and AS/400 269
Formats: Mainframe and AS/400 273
Transports: Mainframe and AS/400 273
Media: Mainframe and AS/400 275
Conclusion 279
IV Network Design, Implementation,
and Management
14 Design and Implementation 283
Top-Down Design Techniques 283
Using NetPICTs to Help
with Systems Integration 285
Wiring Strategies 288
Conclusion 312
1 5 Management and Troubleshooting 31 3
Configuration Management 313
Network Management 315
Troubleshooting 318
Performance 330
Security 335
Conclusion 336
16 The Future of Macintosh Networking 337
Glossary of Networking Terminology 341
A NetPICT Encyclopedia 369
B Listing of Macintosh Networking Companies 383
Index
395
Introduction
r. Richard Feynman (see figure 1) was one of
America's leading nuclear physicists. He
achieved this acclaim by demonstrating his
genius at M.I.T. and Princeton University, and
later by playing a key role in the Manhattan
Project, which produced the first atomic bomb at
Los Alamos, New Mexico.
Figure 1
R.P. feynman.
After World War II, Feynman became a professor at Cornell Uni-
versity, where he tackled the complexities of the newly emerging
science of quantum mechanics. At the time, quantum physics was
awash in competing theories, none of which effectively explained
Live Wired
the intricacies of the sub-atomic world or satisfied the members of
the physics community.
In the late 1940s, Feynman published a sequence of papers that
completely redefined physics. While Feynman’s writings were
revolutionary, they were also full of simple examples and meta-
phors that made the complex subject accessible to a new genera-
tion of particle physicists.
One of the metaphors employed by Fejmman was a simple dia-
gram that depicted particle interactions (see figure 2). In fact, these
“Feynman diagrams” were just a part of an entire language in-
vented by Feynman — a language that has a complete and consis-
tent grammar, with corresponding syntactical rules.
Feynman Diagram
During the remainder of his years, Feynman continued his
achievements: winning the Nobel Prize for Physics in 1965, and
serving on the commission to investigate the Space Shuttle Chal-
lenger disaster.
Richard Feynman died in 1987 after a long batde with cancer, but
his diagrams lived on and are now the lingua franca of the particle
physics community. The symbols gave physicists the power to
express complex phenomena in an elegant, graphical form.
The power of symbolic representation has been repeatedly demon-
strated throughout history. From our numbers and alphabet to the
Figure 2
Feynman Diaoram.
e
Introduction
iconic Macintosh user-interface, s>Tnbolic representation gives us
the power to master complex concepts.
When we master complex concepts through using symbolic
representation, we use a higher level of abstraction to deal with
objects and events. For the programmer, it’s the difference be-
tween writing a computer program in the computer's native
language of assembler or in a higher level s^mibolic language such
as Pascal or HyperTalk.
Just as the Macintosh made computing accessible to the general
public, it also has made computer networking commonplace.
Tasks that once were left to engineers and “authorized” network-
ing professionals are becoming the responsibility of desktop
publishers, architects, and other Macintosh end users.
At the simplest level — networking a Macintosh to a LaserWriter, for
example — Macintosh networking is a plug-and-play operation
that’s as easy as connecting stereo components. But networks are
not always this simple, and for many Macintosh users networking
seems as complex as nuclear physics.
To extend beyond the simple LocalTalk network requires an
understanding of basic networking principles and concepts. When
the average Macintosh user tries to learn and understand these
networking concepts, he or she is often confronted vAth confusing
and ambiguous terms.
This book demystifies Macintosh networking and makes it under-
standable for everyone. Macintosh users, network administrators,
and systems integrators will find the material presented in this
book as clear and understandable as the Macintosh itself. It will
bring an order and a symbolic representation to Macintosh net-
working by introducing a graphical language and diagramming
technique that unifies and simplifies Macintosh network design
o
Live Wired
and systems integration. This language is used throughout the
book to introduce and define complex network protocols and
components.
A disk containing a complete library of this symbolic networking
language is included with this book. The disk contains hundreds of
PICT-based networking symbols, called NetPICTs, that define all
major networking components, computers, and software environ-
ments. These symbols also will be offered by Macintosh network-
ing vendors and will be posted on online services such as
CompuServe, AppleLink, and America Online.
Every hardware or software element that has some relevance to
networking will be explained and diagrammed so that you can
understand its function and role in tlie larger picture. You'll be able
to use your favorite Macintosh drawing program to open these
PICT symbols, so you easily can combine them to solve problems
or diagram your Macintosh network.
The language breaks down Macintosh networking into four basic
categories, or layers. Thoughout the book, obscure networking
concepts like bridging, routing, and tunneling are clearly posi-
tioned and explained. Is Ethernet a networking protocol or cable?
Is AppleTalk hardware or software? Is it necessary to use the same
networking protocols and cabling on each netw'orked computer?
These are the kind of questions that this book readily answers.
You’ll learn from the diagrams that AppleTalk, Apple’s networking
software, not only runs on Macs and LaserWriters, but also on PCs,
Unix workstations, and many other computers.
This book also provides a practical guide to designing, implement-
ing and troubleshooting AppleTalk and multivendor networks.
Numerous real-world examples are used, so it’s likely that the
“blueprint” of your present (or planned) network exists between
these covers.
o
Introduction
Unlike other computer books that focus solely on products and
have a limited lifetime, this book will focus on making the basic
Apple networking concepts clear and understandable. This book
will be as relevant five years from now as it is today.
Years ago, Feynman’s diagrams profoundly changed the world of
physics by introducing a new graphical language that made
difficult and obscure concepts understandable. For me, the goal of
this book is much more pedestrian than Feynman’s achievements:
to pro\ade a graphical user interface for Macintosh networking,
and to make Mac networking understandable and accessible for
the rest of us.
Jim Anders
March, 1993
e
srstandable to everyone
“One”
Part I covers the fundamentals oi
networking, using concepts and
Part I covers the fundamentals of networking,
metaphors that are understandal
using concepts and metaphors that are
everyone. It establishes the founc
understandable to everyone. It establishes the
that will be used throughout the I
foundation that will be used throughout the book
describe Macintosh networking.
to describe Macintosh networking.
Networking
v/ie.s the loundmion that will
hroughout the book to
Macintosh networking.
1
How Does
Communication
Take Place?
Ithough computer networking appears to be a
_ confusing world of terminology and buzzwords,
the basic concepts can be reduced to four basic,
|H interrelated layers. These layers apply to com-
puter networking, spoken English, written
German, sign language, smoke signals — any
language you choose. This chapter describes each
of these four layers, common to all communication, by starting at
the top with the first layer (the Idea layer) and then proceeding
down through the remaining three layers.
To make the position of the layers easy to remember, the number
of lines in the ends of the layer diagram indicate the number of the
layer. The Idea Layer (layer 1) has one line on its ends; the Expres-
sion Layer (layer 2) has two lines; the Transport Layer (layer 3) has
three; and the Medium Layer (layer 4) has four (see figure 1.1).
These end treatments will gain additional significance and will be
used in other ways later in the book.
Part One Networking FundamentaJs
Figure 1.1
Ilielourluniiaiiienial
layers of coimnunicaiion.
Idea
Expression
Transport
Medium
Idea
The Idea is the reason behind the communication. It’s the thought
or essence behind the message. It’s an abstract concept that, when
expressed, transforms the idea into a specific form and causes the
process to move down into the next layer. Expression.
For example, if you have an idea that you would like to express,
it first starts as an abstract concept in your mind. Fortunately,
developing and fostering these concepts is what our minds do best.
Within your mind you can entertain such abstractions as the fall of
communism, the smell of morning coffee, your expectations of
future Macintosh computers, and the sound of your favorite piece
of music. All communications start at this abstract layer of the
Idea.
Expression
Until ESP becomes a proven and reliable way to communicate, the
external expression of ideas is necessary in order to prepare the
ideas for conveyance. To express the idea of the fall of commu-
nism, you could use English, Braille, or a sequence of pictures.
Each format of expression has inherent advantages and disadvan-
tages that make one format better than others for a particular
situation.
Chapter One How Does Communication Take Place?
Transport
Once the abstract idea has been expressed in an external form, the
message then can be transported from the origin to the destina-
tion. This is the function of the Transport layer. Each method of
transport has certain rules, or protocols, that must be adhered to in
order to ensure error-free delivery of the message.
Just as there are many different forms of expression, there also are
many different transport mechanisms. If 1 decide to tell you about
the fall of communism by speaking to you direcdy, 1 am using a
specific, mutually agreeable protocol known as conversation.
Usually, a conversation is a two-way process. One person speaks;
another person listens and acknowledges the message. The order
then reverses and the process repeats itself.
I could also convey my English-based message about the fall of
communism using a written transport protocol. The rules for
conveying written English are much different than those for
spoken English. To convey ideas about the fall of communism
using the written transport protocols of English, I write on a piece
of paper, starting at the top, writing from left to right. I then convey
the message by handing you the piece of paper.
Of course, the idea and the expression (the language) remain the
same using either transport mechanism. The only thing that has
changed is the method of transport.
Medium
The last step in the process is the communications Medium — the
mechanism used by the Transport layer to deliver the message.
o
Part One Networking Fundamentals
Figure 1.2
Spoken tomniunicaiion
uses compiessed air.
When you deliver a message by speaking, the delivery mechanism
is compressed air. Your lungs force compressed air over your vocal
cords, which vibrate and in turn vibrate the air and create modu-
lated waves of compressed air (see figure 1.2). These sound waves
hit the listener’s ear drum, which vibrates, causing electrical
impulses to be sent to the brain where they’re finally "heard.”
NOTE: So, just in case you were wondering; if a tree falls in the forest and
there is no one there, it doesn’t make a noise, it just compresses the air.
If I want to deliver my hand-written document on the fall of
communism, I can choose from numerous delivery media — hand
delivery, mail, fax, or Federal Express. Each method has distinct
benefits and disadvantages, but they all do the same thing: they
transmit my hand-written document.
o
Chapter One How Does Communication Take Place?
Matching Layers
By using separate four-layer diagrams for the sender and the
recipient of the message, we can better illustrate the degree of
interaction between the two parties (see figure 1.3).
Speaker
Listener
Figure 1 .3
conversaiion.
In order for communications to be successful, each corresponding
layer of the sender and the recipient must match. To illustrate this
point, let’s analyze a communications transaction from the bottom
up.
The sender’s and the recipient’s chosen communications medium
(layer 4) is compressed air. For this example, we’ll assume that
they’re within speaking distance of each other in normal atmo-
spheric conditions, and therefore would have no problem commu-
nicating over this medium. If either of the participants were deaf,
then a better choice for the communication medium might be
visual in nature.
Both parties are familiar with the protocol of spoken language, so
the next layer up the stack, the Transport layer (layer 3), matches as
well. If one of the participants only understood the protocol of
spoken language and the other only understood the reading of lips,
the layers would not match and communication would be unlikely.
o
Part One Networking Fundamentals
Moving up to the Expression layer (layer 2), both parties under-
stand the English format. Here, as with the other layers, if there
was a mismatch where one person understood English and the
other only French, the communication process would not occur.
Finally, even at the uppermost Idea layer (layer 1), there must be a
match between the sender and the receiver. If the recipient of the
message has no idea what communism is, or was, the purpose of
the message will be lost. The beauty of this diagramming tech-
nique is that all forms and instances of communications can be
diagrammed and analyzed.
What if the sender and recipient are too far away for the medium of
compressed air to work effectively? Let's say they were 100 yards
apart. In this case, the medium would have to be altered in order to
support the greater distance (see figure 1.4).
Figure 1.4
II ihe medium of
compiessedairis
losuilicient perhaps
radio waves will work.
A walkie-talkie is such a device. It converts the compressed air
waves into electrical signals. These signals are transmitted as radio
waves to another walkie-talkie, which reverses the process and
o
Chapter One How Does Communication Take Place?
converts tlie radio waves back into electrical signals. These electri-
cal signals vibrate the internal speaker, which in turn compresses
the air so the listener can hear the message (see figure 1.5).
Speaker
Figure 1.5
Diaofaniining a complex
commonicaiion using
walkie-talkies.
Notice how each side of the diagram matches the other. The
compressed air medium of the sender’s mouth matches the
medium of the microphone; the compressed air medium of the
walkie-talkie’s speaker matches the medium of the receiver’s ear.
This conversion process only happens at the Medium layer (layer
4) since it’s strictly a medium conversion. Walkie-talkies are
ignorant of the protocols of spoken communication and therefore
the process is limited to the medium layer. You’re totally free to be
rude, incoherent, and ignore all the established protocols of
spoken communication. Since the walkie-talkie only operates at
the lowest layer, it will not attempt to police the conversation.
o
Part One Networking Fundamentals
Conclusion
Each layer of the communication process serves a specific func-
tion, but provides us with unlimited choice. We can tailor the
method of expression, transport, and medium to best convey our
ideas. Each layer is connected to its neighbor, but each layer can
also stand independently and can be analyzed and judged on its
own merits. If English doesn’t properly convey a certain idea, then
perhaps haiku can do a better job. If the U.S. Postal Service is too
slow, then perhaps Federal Express is a better choice.
A key tenet of this idea is that when two or more parties are in-
volved, communication can only occur when their corresponding
layers match identically.
How Does
Networking Take
Place?
his chapter introduces computer networking by
defining network concepts in terms of the four
basic concepts of communication (Idea, Expres-
sion, Transport, and Medium) as defined in the
first chapter.
As with any communication, computer network-
ing can be broken down into four layers (see figure
2.1). We’ll use slightly different nomenclature (nomenclature that’s
specific to networking) to describe these layers.
Services
Format
Protocol
Cabling
Figure 2.1
liielourfunilanienial
layers Dimpuiei
neiwoiking.
Part One Networking Fundamentals
Figure 2.2
Clieni/Servei Willi Mac
anillaseiWrlier.
Idea > Services and Programs
In Chapter 1, the first layer represented the idea — the reason
behind the communication. In general, layer one embodies the
purpose behind the communication. With computer networking,
the purpose or reason behind the communication is to deliver
services to the participants.
In a simple network consisting of a Macintosh and a LaserWriter,
the LaserWriter provides a print service to the Macintosh user (see
figure 2.2). Software on the Macintosh communicates with soft-
ware on the LaserWriter and establishes what is known as a client/
server relationship.
Macintosh Client LaserWriter Server
The client/server model has been used throughout all aspects of
Macintosh networking to provide access to file, database, and mail
servers. Today, the concept continues to expand to include
workgroup applications such as calendar and time management
programs, document librarian systems, and System 7’s Publish and
Subscribe capabilities.
o
Chapter Two How Does Networking Take Place?
Expression > Formats
Humans rely on standard languages to provide effective communi-
cations. We have numerous expressive formats to satisfy different
situations and cultures. The same is true for computers. Informa-
tion can be represented in thousands of computer formats. Some
of these formats are widely adopted standards, while most of them
are unique to a specific computer application.
To start with, all computer formats have at least one thing in
common — they’re all comprised of binary data. That means that
all computer file formats consist of a sequence of ones and zeros.
Each one and zero is called a btf of information. For convenience,
these ones and zeros are often put into eight-bit groups, com-
monly called bytes (see figure 2.3).
00
CM CMtO
T- CO f- 00 tr CM T- —
a] 010100 ( 0 ]—
10100001
00010110
11111111
00001101 —
Each column of a binary
number Is a successive
power of 2 ...
Each one and zero Is
a bit of Information...
Each group of eight
Is a called a byte or
octet.
This byte represents the
decimal value 13 (8+4+1)
and In the ASCII code,
represents a Carriage
Retura
Figure 2.3
Oils anil bytes.
Octets
Bytes are often referred to as octets, particularly in the networking community.
o
Part One Networking Fundamentals
Early in the history of computing, it became obvious that digital
information would need to be standardized. For example, ASCII
(American Standard Code for Information Interchange) has been a
widely used and popular standard for some time. It uses seven bits
of information to represent most of the characters on an English
typewriter keyboard.
Within a byte of information, there are 256 possible characters,
ranging from 00000000 to 11111111. Of these, ASCII uses the first
128 characters to represent common alphanumeric characters and
special characters unique to data processing. The first 32 charac-
ters of the ASCII code are special control characters that cannot be
printed: these control characters are used to represent characters
such as carriage returns (which have an ASCII decimal value of 13
and a binary value of 00001101), line feeds, and tabs. The remain-
ing 128 values are used for special characters or symbols. The
Greek letter “pi,” for example, has an ASCII decimal value of 185 an
a binary valueoflOlllOOl.
The problem with ASCII is that it doesn’t have enough values to
encode all the symbols used throughout the world. To accommo-
date other alphabets and symbols, a more expansive standard was
needed.
To solve this problem, industry groups put forth the Unicode
standard, which uses 16 bits, for 65,536 possible characters. This
standard can accommodate every symbolic character used in the
world today and is destined to eclipse the ASCII standard. Apple
uses double-byte characters (but not yet the Unicode standard) in
System 7.1 to accommodate these additional characters.
Today, many computer applications, particularly word processing
applications, can read and write the ASCII standard. But since
ASCII only encodes information about the characters and basic
formatting, it is not used as a native word processing format.
Chapter TwoHovv Does Networking Take Place?
Therefore, every word processor developer must devise another
special binary format to handle all the other attributes common to
modern word processors. With an application such as Microsoft
Word for the Macintosh, these attributes can include color, fonts,
graphics (of varying formats), and even sound. Of course, when
you save a document that has tliese elements as an ASCII File, all
that remains is the plaintext. The other elements are lost.
There are many other standard formats, such as PostScript,
QuickDraw, PICT, TIFF, GIF, EBCDIC, QuickTime, DXF, and IGES.
For the most part, these formats are independent of the computer
on which they were created. A QuickTime file, composed of a
specific sequence of ones and zeros, can be played on a Macintosh,
a PC, or a Silicon Graphics workstation. All that is needed is an
application that understands the QuickTime format.
Some formats use other formats as a starting point. For example,
PostScript, DXF, and IGES Files are composed of strings of ASCII
text. The ASCII text is actually a series of commands or statements
in the higher language. (The English format is based upon the
Roman alphabet text format, as is the French format and the
Italian format.)
When it comes to the native format used by most applications,
nearly every one has its own unique (and often proprietary) file
format. In other words, if you create identical documents in
MacWrite and Microsoft Word, each application would use differ-
ent patterns of ones and zeros to describe the same data. Both
applications can import and export each other’s format, but that’s
because the respective applicadons can translate the competitor’s
format.
More than any other layer, the Format layer is responsible for
successful communications. A transcontinental telephone conver-
sation is an apt analogy. It’s easy to dial the phone number of
o
Part One Networking Fundamentals
someone who lives in France. The call gets routed and placed widi
all kinds of sophisticated technology, but the ultimate measure of
successful communications is whether or not I can understand
French, or if my counterpart understands English (see figure 2.4).
Figure 2.4
Hieeipressionlor
foimai) of ilie message is
crucial lor uodersianding.
Yo! What’s
for breakfast?
Omelette....
Du Fromage!
vr~
Establishing a common format for communications is by far the
most difficult challenge in computer networking. It’s the task that
requires the most thought and effort to solve.
Transport > Protocols
Until this point, all we’ve done is establish a service and agree on a
common descriptive format. No information has moved across the
network. The rules for moving formatted information across the
network are the function of the third level; protocols.
Chapter Two How Does NeuvorkingTake Place?
Networking software must solve many problems: the rules that
describe the solutions are the networking protocols. First, the
sender and recipient must be uniquely identified on the network.
(This is analogous to assigning phone numbers to identify
phones — you need to know the number of the phone you want to
call.) Next, there must be a means to route data over complex
pathways without the loss of data. This is just a sample of the
functions of the Protocol layer.
Apple developed a protocol known as AppleTalk that very elegantly
solved the problem of identifying the sender and the recipient.
AppleTalk provides reliable, error-free transmissions, and chooses
appropriate routes on the netu'ork. AppleTalk, like other network-
ing protocols (such as TCP/IP, Novell, and DECnet), exists only in
software. All Macintosh computers and most of Apple’s networked
printers come equipped with AppleTalk software.
Although AppleTalk is part of the Macintosh System software, it is
by no means limited to the Macintosh. The AppleTalk protocols
are available for many other computers and networking devices
(see figure 2.5). DOS-equipped PCs can use Farallon’s PhoneNet
Talk PC software to gain the AppleTalk protocols. PCs running
OS/2 2.0 come with AppleTalk protocols. (This was one of the goals
of the Apple/IBM alliance.) AppleTalk protocols are available for
other computer platforms, including NeXT workstations running
NextStep 2.0 and Digital’s VAX family.
Computers — the Macintosh included — are not limited to running a
single networking protocol. Just as humans can deal with multiple
transport methods, so can computers.
Your Macintosh can have several transport protocols loaded at one
time (see figure 2.6). In addition to the standard AppleTalk proto-
col, you could install MacTCP from Apple and DECnet for
Macintosh from Digital. These protocols would enable your
Part OneNetworking Fundamentals
Figure 2.5
Applelalk protocols on
other platforms.
Figure 2.6
A Macintosh roiinlnp
multiple protocols.
Macintosh to communicate using the native transport protocol of
other computers.
Intel/DOS Machine
Services
“Chooser”
1
DEC VAX
PostScript
AFP Server & |
PAP Spooler
NeXT Workstation
AppleTalk /
AFP & PostScript
AFP Client
^ LocalTalk or [
Ethernet 1
AppleTalk /
PostScript
Ethernet [
AppleTalk
Ethernet
Macintosh
Services
PostScript
AppleTalk
DECnet TCP/IP
Ethernet
MacTCP equips the Macintosh with the TCP/IP protocol used by
most UNIX computers. Ruiming DECnet for Macintosh turns your
Mac into a DECnet node and enables it to communicate with
DECnet-equipped computers such as PCs and VAXes.
Figure 2.7 shows another diagram depicting the concept of mul-
tiple protocols. Here, the protocols are shown as geometric shapes.
Those computers that are able to send and receive these protocols
have corresponding notches designed to handle the protocols.
o
Chapter Two How Does Networking Take Place?
Computer A Computer B Computer C Computer D
Different Networking Protocols
(I.e. AppleTalk, DECnet, TCP/IP)
O □ A
Figure 2.7
Multiple piolocols.
While it’s possible for a Macintosh to run multiple transport
protocols, it’s important to note that these transports don’t always
support every Macintosh service. For example, the Chooser only
works with AppleTalk transport protocol. The file service of NFS is
only accessible with the TCP/IP transport.
There are only a few Macintosh services that work with multiple
transport protocols. The best example of such a service is MacX,
which works over several transports, such as AppleTalk, TCP/IP,
and DECnet. In this case, the choice of a transport depends largely
on the transport protocol of other network participants.
Medium > Cabling
At some point, the networking software, or protocols, must deal
with the physical world to send the message. When we speak, it's
the air that carries our words. When computers speak, they rely on
electromagnetic signals to carr>' the sequences of ones and zeros.
These encode the networking protocols that contain the formatted
information which ultimately delivers the service to the user. There
are several ways computers can transrnit electromagnetic signals,
but the most common way is through a cable.
o
Part One Networking Fundamentals
Every Macintosh has two serial ports. They are most commonly
known as the printer port and the modem port. Both ports can be
used for serial communications, but only the printer port can be
used to connect to Apple’s low-cost netw'orking cabling medium
known as LocalTalk.
Since LocalTalk hardware is included on every Macintosh and on
most Apple printers, it is a simple task to create a basic network.
Just acquire the LocalTalk connectors and connect them using the
appropriate cabling.
Apple’s LocalTalk was designed in the mid 1980s, when Ethernet
hardware cost $750 to $1000 per node. Apple needed an inexpen-
sive way to send AppleTalk protocols over a simple network. At $50
per node, LocalTalk networks rapidly took hold. To get the price
down, Apple made some reasonable tradeoffs.
First, LocalTalk networks have a limited bandwidth, of 230,400 bits
per second (230.4 Kbps). This limits the maximum number of
connected nodes to 32. This was much less than Etliernet or other
cabling systems, but it was reasonable for Apple’s small workgroup
strategy of 1985. While LocalTalk put serious limitations on large
and complex networks, Apple took the necessary steps to expand
the AppleTalk protocols to other popular cabling systems.
First came support for Ethernet. Apple and third-party vendors
began to develop Ethernet cards for the Macintosh. Included with
the cards was special driver software that enabled the AppleTalk
netw’orking software to communicate over the Ethernet card,
instead of the standard LocalTalk port.
Apple decided to trademark the software that sends AppleTalk
protocols over Ethernet, and called it EtherTalk. But don’t be
confused. It’s not a new kind of cable or a different protocol. Just
think of EtherTalk as AppleTalk protocols over Ethernet cabling,
and it will make more sense.
Chapter Two How Does Networking Take Place?
Ethernet has a much higher bandwidth than LocalTalk. It’s capable
of carrying information at a rate of 10 million bits per second (10
Mbps). This is roughly 40 times the bandwidth of LocalTalk. There
are three different implementations of Ethernet: all have the same
bandwidth, but the cable type, connectors, and maximum segment
lengths differ.
After Ethernet, Apple provided support for Token Ring networks.
Token Ring is common in the IBM mainframe and minicomputer
world. There are two implementations: a 4 Mbps version, and a 16
Mbps version. As with Ethernet, Apple included driver software to
connect the higher-level AppleTalk protocols to the Token Ring
hardware. The Apple trademark for this is TokenTalk. Again, just
think of it as AppleTalk protocols running over Token Ring cabling.
Usually, the decision to use Token Ring as a cabling medium has
more to do with IBM connectivity than it has to do with the relative
merits of the cabling system. Token Ring cards are significantly
more expensive than Ethernet cards, but as vve’U see later on.
Token Ring cabling operates in a fundamentally different way than
Ethernet does, and it might make sense to consider Token Ring
even if you don’t have an IBM mainframe.
More recently, Apple introduced AppleTalk support for another
kind of cabling. With AppleTalk Remote Access Protocol, or ARAP,
AppleTalk protocols are sent over serial connections, such as RS-
232 cabling used by modems. As with the other cabling choices,
only the low-level drivers are affected. The higher-level AppleTalk
protocols remain the same.
ARAP makes it possible for a remote Macintosh client to connect to
a Macintosh or another server by calling in through a modem. The
computer that receives the call acts as the ARAP server. Once the
connection is made, the remote Macintosh client becomes a ptu-t
of the network and is able to access the serxdces as if the remote
o
Part One Networking Fundamentals
Figure 2.8
Remote access diBoiam.
Mac were directly connected to the network. Of course, since the
dial-up modem connection is significantly slower than either
LocalTalk or Ethernet, the trade-off is performance.
Apple’s Remote Access software provides both tlie client and
server applications for the Mac (see figure 2.8). However, third-
party vendors are offering ARA servers that run on dedicated server
boxes, such as Shiva’s LanRover products and Computer Methods
Corporation’s AsyncServeR (which enables a DEC VAX to become
an enterprise-wide AppleTalk Remote Access server).
Remote Macintosh
Apple has already released the specification for the next generation
of high-speed cabling systems. Fiber Distributed Data Intercon-
nect (or FDDI) has a bandwidth that is an order of magnitude
more than Etliernet (see figure 2.9). At 100 Mbps, FDDI interfaces
should provide the necessary bandwidth for the demanding
applications where voice data and real-time video are sent over the
Chapter Two How Does Networking Take Place?
network. Several vendors are already starting to offer FDDl NuBus
cards for the Macintosh, and Apple has no doubt trademarked the
FDDITalk name. Prices are still high, but just as Ethernet prices
have plummeted during the last few years, you can expect the
same to happen with FDDI technology as well.
Electromagnetic communication is not limited to copper wiring or
fiber optics. In fact, it seems likely that the next generation of small
hand-held computers, as exemplified by Apple’s Newton, will not
use any cabling.
Instead, different portions of the electromagnetic spectrum v\all be
used in place of physical cabling. Your TV or VCR remote control
uses infrared waves to communicate. Your pager, remote auto
alarm, and cellular phone use radio waves to communicate. All
these methods will be common communications and networking
mediums in the near future.
Figure 2.9
FDOI.
Part One Networking Fundamentals
Conclusion
Networking may appear to be a a complex subject, but it easily can
be broken down into four layers. Services are the applications,
which use a variety of formats or data structures. These formats are
delivered with networking transport protocols, over some kind of
delivery medium, such as cabling. This is true of all computers and
network systems. Once this concept is understood, all that remains
is to determine which buzzwords belong to which layer.
Network
Diagramming
with NetPICTs
n this chapter, the symbolic language based
on the NetPICT symbols and the associated
diagramming technique is introduced and
explained, using several basic Macintosh
network scenarios. Using the four-layer
NetPICT symbol as a starting point, the
seven-layer OSI reference model is finally
explained in an easily understandable fashion.
Introducing the NetPICT symbol
To this point, the NetPICT symbols have been used to describe
fairly simple networking scenarios. Combined with a few simple
rules, we will use the symbols to depict common Macintosh
networking scenarios, and to solve some simple problems.
Part One Networking Fundamentals
Let’s start with a typical Macintosh. Out of the box, the Macintosh,
combined with its System softwcU'e, fulfills each of the four layers.
Starting at the top, the most familiar Macintosh service is the
Chooser desk accessory. Found under the Apple menu, the
Chooser application is a service that delivers other services to the
user (see figure 3.1). With the Chooser, a Macintosh user can select
services from the network, including file, print, mail, and other
services.
Figure 3.1
HieslandardMac
Service layei.
Macintosh Client
Once a service (a LaserWriter, for example) has been selected,
other Macintosh applications, such as MacDraw Pro, can utilize
that service and print documents. So, in addition to the Chooser,
we’ve also placed MacDraw Pro as a typical application that
appears at the Service layer.
Working our way down, the next layer is the Format layer (see
figure 3.2). Here, we see that the MacDraw Pro application sup-
ports several different formats.
First, there is the native binary MacDraw Pro format that is unique
to MacDraw Pro. This is the format that writes to your disk when
you save a MacDraw Pro document.
Another format used by MacDraw Pro is QuickDraw, which is used
as an imaging format to draw entities on the Macintosh screen.
Chapter Three Netw^ork Diagramming with NetPICTs
It also can be used for printing, when combined with a QuickDraw
printer such as Apple’s StyleWriter. If you’re using a PostScript
printer instead, the MacDraw Pro application generates yet an-
other format, known as PostScript.
Macintosh Client
Figure 3.2
llie standard Mac
Format layer.
As mentioned before, every Macintosh comes equipped with
AppleTalk protocols as part of its System software. Thus, we’ve
placed AppleTalk in the third Protocol layer (see figure 3.3). The
PostScript generated in the second layer is now passed down to the
third level for subsequent delivery.
Macintosh Client
Chooser
MacDraw Pro
MacDraw Binary
QuickDraw
PostScript
AppleTalk
Figure 3.3
IhesiandBidMac
Pioiocol layer.
And finally at the fourth layer, the LocalTalk cabling, which pro-
vides the physical cabling connection to the outside world, is used
to actually transmit the PostScript data, along with network and
printer control data, to the printer (see figure 3.4).
Part One Networking Fundamentals
Figure 3.4
Ihe Standard Mac
Cabling layer.
Macintosh Client
Chooser
MacDraw Pro
MacDraw Binary
QuickDraw
PostScript
AppleTalk
LocalTalk
Figure 3.5
Ihe standard PC
Service layer.
To further illustrate the function of each layer, let's look at a few
examples of other systems. With an IBM (or compatible) PC, the
layers perform the same function.
At the Semce layer, the PC user chooses services in different ways,
often depending on the application (see figure 3.5). Printers are
usually assigned code names such as LPTl or LPT2. With DOS, the
printer selection is usually typed in as part of the command line.
With Windows-equipped PCs, the printers often have iconic
representations.
PC Client
As witli the Macintosh example, applications also belong at the
service layer. So Lotus 1-2-3 (and any other application) would be
placed here as well.
Lotus 1-2-3 supports a number of formats, in addition to its own
native file format. Some of these formats are used for printing.
o
Chapter Three Network Diagramming with NetPICTs
These include Hewlett-Packard’s PCL, used by LaserJet printers,
and the industry' standard PostScript, which is understood by
many laser printers.
PC Client
At the transport layer, the PC user is confronted with many confus-
ing choices (see figure 3.7). Unlike the Macintosh, the PC was
never designed with a networking protocol built-in, so third-party
companies proceeded to fill the void. Novell NetWare, Banyan
Vines, and Microsoft LAN Manager are all examples of popular
transport protocols available for the PC.
PC Client
Figure 3.6
IhesianMPC
loimai layer.
Figures.?
HiesiandardPC
Proiocol layer.
There's a similar problem at the lowest Cabling/Media layer as well
(see figure 3.8). The only communications port that comes stan-
dard on a PC is a serial communications port. Serial ports are not
normally thought of as networking ports, since they only support a
single connection at one time. To provide a PC with a true network
connection, a networking card must be purchased.
Part One Networking Fundamentals
Figure 3.8
Ilie standard PC
Cabling layer.
PC Client
Lotus Native Format
HP PCL
PostScript
Novell Netware
Banyan Vines
MS LAN Manager
Arcnet
Token Ring
Ethernet
There are several networking cards that are popular for PCs. Today,
most PC networking cards are either Token Ring, Ethernet, and
Arcnet. In its day, Arcnet was the LocalTalk of PCs. It was inexpen-
sive and used standard twisted pair wiring. But as the price of
Ethernet and Token Ring cards dropped over the years, these cards
have come to dominate the PC market.
Swapping Layers
Let’s return to the Macintosh example used earlier. Here we have a
Macintosh that “speaks” AppleTalk over a LocalTalk interface. If
we wanted to trade-up to an Ethernet connection, we would
simply install a networking card and install the appropriate
EtherTalk driver software (figure 3.9).
Figure 3.9
Swapping locallalk lor
Elheinei.
Macintosh Client
Macintosh Client
Chooser
MacDraw Pro
Chooser
MacDraw Pro
MacDraw Binary
QuickDraw
PostScript
\ / MacDraw Binary
) ( QuickDraw
/ \ PostScript
AppleTalk
1 1 AppleTalk
> LocalTalk
c > Ethernet -
Chapter Three Network Diagramming with NetPICTs
As far as the diagram is concerned, the only change we would need
to make is to replace the current Cabling layer of LocalTalk with
one of Ethernet. All the other layers remain the same. With the PC,
we might want to get a bit more ambitious. Here, we might want to
get rid of the Novell networking protocols and replace them with
the AppleTalk protocols (see figure 3.10). This can be done with
Farallon’s PhoneNet PC. PhoneNet PC works with most PC
Ethernet cards, or with PC LocalTalk cards that are offered by
Farallon and other third-party vendors.
PC Client
PC Client
LPT1
Lotus 1-2-3
LPT1
Lotus 1-2-3
Lotus Native Format \ /
HP PCL ) (
PostScript / \
Lotus Native Format
HP PCL
PostScript
AppleTalk J [
AppleTalk
LocalTalk < >
Ethernet -
Figure 3.10
PC ninoApplelalk Willi
locallalkanilEiliernet
As evidenced from the prior example, the process of network
design or systems integration often involves the swapping and
substitution of various layers, in order to achieve the desired
results.
IMPORTANT NOTE: The most important rule to remember is: Two
computers can only fully communicate when each of their respective four
layers match exactly, or have counterparts that can work together.
Matching Layers
Let's look at the NetPICTs of a Macintosh and LaserWriter and see
why they're able to communicate. As shown in figure 3.1 1, the
Macintosh and the LaserWriter match all four layers.
o
Part One Networking Fundamentals
Figure 3.1 1
MacandlaserWriiei
matcliiag layers.
Macintosh Client LaserWriter Server
The LaserWriter icon in the Chooser represents the client portion
that is responsible for printing. It matches and connects with the
server application that runs on the LaserWriter.
At the Format layer, the PostScript generated by MacDraw Pro
can be readily processed by the PostScript interpreter on the
LaserWriter.
Both machines can speak and understand AppleTalk, which is
being used to convey the printing instructions at the Protocol
layer.
Since both machines have LocalTalk interfaces at the Cabling
layer, they can use a common mechanism for carrying the
AppleTalk protocols.
If any one or more of these layers didn't match, the printing
process wouldn’t be possible. For example, Hewlett-Packard offers
a LocalTalk card for certain models of their LaserJet printers. They
also provide AppleTalk protocol and PostScript support as well. If I
were to just install the LocalTalk card into a LaserJet, without also
installing the AppleTalk and PostScript software, I would be unable
to print (see figure 3.12). The same situation would also exist if I
installed the AppleTalk protocols without the PostScript, or vice
versa.
Again, all four layers must match exacdy for communications to
occur.
Chapter Three Network Diagramming with NetPICTs
Macintosh Client HP LaserJet
Figure 3.12
MacandlaseiJeinoi
matcliinolayeis.
Clients and Servers
Client/server computing is not a product or a specific application.
It’s a way of designing computer programs. It’s a design philoso-
phy. In the client/server model, computers and programs interact
in a way that’s ideal for a networked environment.
The general idea is simple. A client program submits a request to a
server program, which acts on the request along with die requests
of other clients. The obvious example is a file server that handles
the requests of clients that desire file ser\'ices. Database servers,
such as Apple’s Data Access Language (DAL), performs a similar
function for clients desiring database services.
The client/server model even applies to the design of computer
programs. The popular symbolic math program, Mathematica, was
written using a client/server model. When Mathematica is running
on a single computer, such as the Mac, the client and server
portions of the program are running on the same machine. In a
networked environment, a different sei^ver could be used. By
having the option to choose different Mathematica servers, the
user has the option of selecting a ser\'er that runs on a super
minicomputer, while still using the client interface on the Mac.
Throughout this book, the concept of client/server computing will
appear quite often. It will be used to describe the way applica-
tions work together, or how one network layer interacts with its
neighbor.
Part One Networking Fundamentals
Hgure 3.13
OSIReiefence Model with
seven layers.
The OSI Reference Model
and NetPICTs
The four layers used in the NetPICT diagrams are essentially a
condensed version of an industry standard known as the OSI
Reference Model. The model, developed by the International
Standards Organization (ISO), is used to describe the ISO’s stan-
dard networking protocols and also to act as a road map for other
networking protocols as the fundamental layers can be applied
universally.
The OSI Reference Model uses seven layers, as opposed to the four
used by the NetPICT diagrams (see figure 3.13). As we discussed
earlier, the end treatments of the different NetPICT layers serve
different functions.
Application
Presentation
Session
Transport
_JJetwork_
Data Link
Physical
First, the number of lines in the ends of the NetPICT layers indicate
their respective position on the stack. Secondly, many of the end
treatment lines correspond to key layers of the OSI Reference
Model (see figure 3.14).
0
Chapter Three Network Diagramming with NetPICTs
The three lines in the Protocol layer symbolize the three layers in
the OSl Reference Model that focus on the neu\'orking transport
protocol. Another similarity can be found in the fourth Cabling
layer. Here the four line end treatment can be readily divided in
half, where the upper half represents the Data Link OSl layer and
the lower half represents the Physical layer.
The only de\aation from the strict interpretation of the OSl model
is the two lines of the Format layer. Here, the entire Format layer is
represented by the Presentation layer of the OSl model.
For the purposes of this book, however, later we’ll use the two lines
of the Format layer to separate different kinds of formats.
NetPICT OSl Reference Model
Figure 3.14
NelPICIandOSIMalcliino.
So, by starting with our simplified four-layer approach, you already
have a basic understanding of the OSl Model. Table 3.1 describes
the specific relationships between the NetPICT model and the OSl
model. As the function of each layer is described, keep in mind that
each layer dovetails with its neighbor in a client/server relation-
ship.
o
Part One Networking Fundamentals
Table 3.1 OSI Reference Model Layers
OSI Layer
Corresponding
NetPICT layer
Function
Application
Service
The place where network
services and applications
reside; utilizes the formats
established in the next layer
of the stack.
Presentation
Format
These include file formats,
such as PostScript, ASCII
and Microsoft Word; and
file access formats such as
the Apple Filing Protocol.
Session
upper third
of Protocol
Addresses the problem of
establishing and maintain-
ing a connection between
computers; also maintains
a logical sequence to the
communications.
Transport
middle third
of Protocol
Ensures reliable delivery of
tlie message.
Network
bottom third
of Protocol
Essentially addresses the
message for delivery.
Data Link
upper half of
Cabling
Concerns the specific kind
of cabling or communica-
tions medium employed.
Physical
bottom half
of Cabling
The level where the physical
cable or delivery medium
exists.
A few of these descriptions are new and require a brief explana-
tion. The Transport layer ensures that the message is correctly
Chapter Three Network Diagramming with NetPICTs
transmitted. If one portion of the communication transmission is
lost or garbled, it’s the job of the Transport layer to re-transmit the
necessary portion.
The Network layer functions as an envelope containing the mes-
sage. The envelope has the address of the message recipient and
the return address of the sender. These addresses are logical
addresses that are specific to a particular networking protocol. This
logically addressed envelope is known as a datagram. This concept
is very important and will be covered in detail in the next chapter.
The function of the Data Link layer is to place the datagram,
created at the Network layer, inside a network delivery vehicle,
known as a frame. Network delivery frames are specific to the
particular kind of cable being used. An Ethernet frame can hold the
same AppleTalk datagram as a Token Ring frame, but the respec-
tive network frames are different. Another way for Macintosh users
to think of the Data Link layer is by looking at the Mac’s Network
Control Panel. When a Macintosh user chooses a different network
driver with the Network Control Panel, a different Data Link layer
is being selected.
The Physical layer is the level at which the actual delivery medium
is realized. The three different kinds of Ethernet and two different
kinds of Token Ring cable are defined in this layer. The Physical
layer often has its own form of addressing in addition to the
protocol-specific logical addressing. For example, each Ethernet
node, or device, has a unique 48-bit hardware address that remains
constant. LocalTalk, on the other hand, doesn’t use physical
hardware addresses and relies on the logical AppleTalk address
instead. Physical hardware addresses and their relationship to the
protocol logical addresses will be covered in the next chapter.
With that quick overview of the OSI Reference Model, let’s take
another look at how a Macintosh prints to a LaserWriter, but this
time, we’ll use the full seven layers of the OSI Model.
o
Part One Networking Fundamentals
Figure 3.1 5
Priming describeilwilh
lire OSIRelerence Model.
A Seven Layer OSI Example
As before, starting at the top, the most familiar Macintosh Applica-
tion is the Chooser desk accessory. Found under the Apple menu,
the Chooser application delivers other services to the user (see
figure 3.15). With the Chooser, a Macintosh user can select file,
print, mail and otlier services from the network.
Macintosh Client
HP LaserJet
Once a service, such as a LaserWriter print server, has been chosen,
other Macintosh applications, such as MacDraw Pro, can utilize
that service and print documents. So, in addition to the Chooser,
we've also placed MacDraw Pro as an application that appears at
the OSI Application layer.
Working our way down, the next layer is the Presentation layer.
Here, as before, we see that the MacDraw Pro application supports
several different formats. First, there is the native binary MacDraw
Pro format that is unique to MacDraw Pro. This is the format that is
written to your disk when you save a MacDraw Pro document.
Another format used by MacDraw Pro is QuickDraw, which is used
as an imaging format to draw entities on the Macintosh screen. It
also can be used for printing when connected to a QuickDraw
printer such as Apple’s StyleWriter. If you’re printing to a Post-
Script printer instead, the MacDraw Pro application generates yet
another Presentation layer format, known as PostScript.
Chapter ThreeNetwork Diagramming with NetPlCTs
The Presentation layer of PostScript gets sent to the Session layer.
The Session layer is the first layer where the AppleTalk networking
protocols really come into play. At the AppleTalk Session layer,
there’s an Apple protocol known as the Printer Access Protocol
(PAP), pap’s job is to add printer control information to the
PostScript data.
Once the AppleTalk PAP Session layer has done its job, it passes
the data down the stack to the Transport layer. At the Transport
layer, the AppleTalk Transaction Protocol, or ATP, adds transac-
tion-oriented commands to manage the give-and-take nature of
the transmission.
With the transaction data added to the message, it’s placed inside
an AppleTalk datagram at the Network layer. Here, the Datagram
Delivery Protocol (DDP) has the task of creating the datagram and
logically addressing it for delivery.
The addressed datagram is then sent to the Data Link layer where
the selected network driver — either the LocalTalk Link Access
Protocol (LLAP), EtherTalk Link Access Protocol (ELAP), TokenTalk
Link Access Protocol (TLAP), or another AppleTalk network
driver — places the datagram into the appropriate network frame.
Finally, the Data Link layer passes the frame onto the Physical
cabling layer and the message is sent to the recipient.
Conclusion
The seven layer OSI Reference Model provides a complete — but
often complex — view of Macintosh networking. To address this
problem of complexity, a simplified four-layer symbology has
been introduced which essentially collapses some of the detail.
The collapsed layers of the OSI Model have importance and
Part One Networking Fundamentals
significance, and will be discussed, where appropriate, throughout
this book. But, for most basic Macintosh networking activities, the
four layer NetPlCT approach will enhance the reader’s under-
standing and comprehension of a complicated world.
o
Networking
Concepts
e
xtending and adding to the NetPICT concept,
this chapter explores concepts such as network
addressing (botli logical and physical) and how
digital signals are conveyed over a network.
Types of Networks
Basically, there are two kinds of computer networks: circuit-
switched and packet-switched. Circuit-switched networks ermpXoy
point-to-point links between the participating nodes of the net-
work (see figure 4.1). The link is dedicated between the partici-
pants of the connection. This is similar to most telephone systems,
where the phone company uses elaborate switching mechanisms
to connect the callers.
Packet-switched netivorks uiWize a common connection which is
shared simultaneously among all nodes of the network. Members
of a packet-switched network communicate by sending discrete
packets of information that are identified with the network address
of the sender and recipient. To better describe the differences
between circuit and packet switching, let’s imagine a classroom
full of students.
Part OneNetworking Fundamentals
Figure 4.1
Cifcuiiswitcliino.
The students aren’t allowed to talk out loud, so one of them
fashions a circuit-switched network with two tin cans and some
string. Each time it’s used, the cormection is dedicated between
the two participants. The switching occurs when other students
need access to the network. Then, the tin cans get passed along to
another pair of students. Of course, waiting for the tin cans be-
comes tiring, and when a number of students need to communi-
cate, the circuit-switched network becomes inefficient.
So, in place of the tin cans, another solution was devised. The
students began to pass notes to each other. To identify who
receives emd who sends the note, each note has the names of the
sender and the recipient on the outside of the note. This network-
ing technique is known as packet-switching.
If the sending and receiving students are sitting next to each other,
they simply can pass messages between themselves. If the sender
and the recipient are sitting at opposite ends of the classroom, the
Chapter Four Networking Concepts
note has to be passed from student to student. Each student who
handles the note looks at the names of the sender and recipient to
determine how best to route the message. This is similar to the task
that network routers perform when they route network packets
from the source to their destination (see Figure 4.2).
Figure 4.2
Packet swiicliino.
One advantage of the note-passing packet-switching scheme is
that multiple notes or packets can be handled at any given instant.
This is in direct contrast to the circuit-switching technique, where
each connection remains dedicated for the duration of the transac-
tion.
Most LANs, such as LocalTalk, Ethernet, and Token Ring, rely on
packet-switching technology. Obviously, one of the important
aspects of a packet-switched network involves the identification of
the sender and the recipient. This general problem is referred to as
network addressing.
Part One Networking Fundamentals
Network addressing, like the street address of where you live or the
name of a note-passing student, simply means to uniquely identify
the network participant. There are two different kinds of network
addressing: logical and physical.
Logical addressingis software-based. It’s how a specific networking
protocol, such as AppleTalk, identifies the senders, recipients, and
other devices on the network. Physical addressing is hardware-
based. It’s tied to a specific cabling system. Let’s compare these
two different addressing schemes.
Logical Addressing
Every networking protocol, such as AppleTalk, DECnet, and
TCP/IP, has a scheme to identify members of its respective net-
work. Essentially, this means every network member, or node, has
a unique identification number.
With AppleTalk, this identifying number, or node number, is based
on an eight-bit number. With eight bits of information, the maxi-
mum number of nodes is 2^ or 256 (see figure 4.3). For small
networks, having a limit of 256 nodes is not a problem. In fact,
Apple’s LocalTalk cabling system permits no more than 32 nodes.
(With LocalTalk, the restriction is a physical or electrical limit, not
a logical limit of node addresses.)
Figure 4.3
Applelalk node numbers.
22 32
18
73 63
57
But many networks require more than 256 AppleTalk nodes, and
another mechanism was needed to support these larger networks.
So, Apple developed a concept that permitted groups of nodes,
called “networks,” to be established: each AppleTalk network
would have its own range of 256 potential node IDs.
Chapter Four Networking Concepts
To create the boundaries between these different AppleTalk
networks, Apple developed the specifications for a device knovvTi as
an AppleTalk router. AppleTalk routers are the glue that connects
separate AppleTalk networks into a larger whole known as an
internetu>ork. AppleTalk routers separate and define the bound-
aries between AppleTalk networks.
Each AppleTalk network has a unique ID number that is stored
inside each router connected to that network. In the simplest case
of two AppleTalk networks connected by a single router, the router
maintains the AppleTalk network numbers for each of the two
networks. Routers will be discussed in depth later in the book.
The current AppleTalk specifications provide a 16-bit network
number. With 16 bits there can be 2 or 65,536 unique AppleTalk
networks (see figure 4.4). And since each network can potentially
support 256 nodes, AppleTalk can support a theoretical maximum
of 16,777,216 devices!
Figure 4.4
Applelalk network
numbers.
The 16-bit network and 8-bit node number make up only part of a
complete AppleTalk logical address. An additional 8 bits is used to
define something known as an AppleTalk socket number.
o
Part One Networking Fundamentals
Your Macintosh can handle a number of network chores at one
time. At any given instant, a Macintosh could be accessing a file
server, printing to a LaserWriter print server, and receiving an
E-mail message. To keep these diverse transactions separate and
distinct, AppleTalk assigns a unique socket number for every
logical connection. With socket numbers consisting of 8-bit values,
there are 256 possible socket numbers that can be used (see figure
4.5). Socket numbers start at 0 and end at 255. Both 0 and 255 are
undefined. Half of the remaining socket numbers, from 1 to 127,
are reserved for special system use by Apple. The other half, from
128 to 254, are pooled resources available for general use by
applications.
Figure 4.5
Applelalk socket
npoibeis.
Network #1
12
Socket 132 is being used...
1.28.132
(n
r
32
34
57
58
Network #2
JC
.to communicate with
Socket 143.
22
32
18
r
n
T
T
2.63.143
57
When you print a document, a socket number for that transaction
is automatically assigned from the pool. When the job is com-
pleted, the number is returned to the pool for subsequent use.
Collectively, these three numbers — network, node, and socket —
make up a unique AppleTalk logical address. When a Macintosh
and LaserWriter engage in printing transactions, they identify the
o
Chapter Four Networking Concepts
participants using netw'ork, node and socket numbers — for in-
stance, Macintosh node number 22, located in network 100, is
communicating over socket 129 to LaserWriter node 32, in network
101, with socket 130.
The choice of three numbers — the 16-bit network number and the
8-bit node and socket number — is specific to AppleTalk. Other
networking protocols, such as DECnet and TCP/IP, use similar but
different numbering schemes to logically identify their nodes. The
important thing to remember about these logical network ad-
dresses is that they are tied to a specific networking protocol, and
they only exist in software.
Because Macintosh and other computers can often run multiple
networking protocols concurrendy, it’s often common to have
multiple logical addresses on a single computer. For example, a
Macintosh that’s running AppleTalk, DECnet, and TCP/IP concur-
rently would have three different logical addresses (see figure 4.6).
Its AppleTalk address might be 12.22 (network.node), its DECnet
address might be 5.2 (area.node), and its TCP/IP address might be
100.22.128.132.
12.22 AppleTalk
5.2 DECnet
100.22.128.132 TCP/IP
r
n
D
i
Figure 4.6
Oifleieni logical
addresses on one Mac.
Every time this Macintosh sends or receives AppleTalk communi-
cations, its logical network AppleTalk address is 12.22; when
DECnet is sent or received, the logical DECnet address of the
Macintosh is 5.2.
Part One Networking Fundamentals
Figure 4.7
Elhemet physical
addiessing.
Physical Addressing
Another kind of addressing is used often to facilitate the transmis-
sion of data over a physical cabling medium. For example,
Ethernet uses a unique number to identify each Ethernet device
(see figure 4.7). The best example of an Ethernet device is an
Ethernet card. Ethernet relies on a unique 48-bit number, where
the first 24 bits identify the vendor of the device and the last 24 bits
identify the specific device — kind of like a serial number. These
numbers are stored permanently inside the Ethernet device. If you
swap Ethernet cards with another Macintosh, the physical
Ethernet address remains with the card rather than with the
Macintosh.
This Mac has an Ethernet
hardware address of
08 - 22 - 12 - 32 - 37-14
¥ ^
3E
r
n
This Mac has an Ethernet
hardware address of
08 - 22 - 12 - 32 - 12-12
With Ethernet, access to the cable and communications over that
cable are controlled by the physical Ethernet addresses. Each
networking protocol is responsible for equating its protocol-
specific logical address to the Ethernet physical address that is
present.
So, going back to our example of a Macintosh running three
different protocols (AppleTalk, DECnet, and TCP/IP), each
Chapter Four Networking Concepts
protocol has a different logical identification number — but all
must convert their respective logical addresses to the same
Ethernet physical address (see figure 4.8).
12.22
5.2
100 . 22 . 128.132
AppleTalk
DECnet
TCP/IP
This Mac has an Ethernet
hardware address of
08 - 22 - 12 - 32 - 12-12
TCP/IP uses a special protocol called ARP, which stands for
Address Resolution Protocol. It resolves the different logical and
physical addresses. AppleTalk uses a similar protocol called the
AppleTalk Address Resolution Protocol, or AARP. AARP equates
logical AppleTalk addresses with physical Ethernet addresses.
A table containing the AppleTalk addresses, along with their
respective Ethernet addresses, is maintained on each Ethernet-
equipped Macintosh. This table is known as the Address Mapping
Table, or AMT. When a Macintosh needs to send a message to a
specific AppleTalk address, the corresponding Ethernet address is
determined from the AMT. Then, when the message is sent to the
Cabling layer, Ethernet addressing is used to deliver the goods.
Figure 4.8
lllieinet physical
addressing and muliiple
logical addresses.
Part One Networking Fundamentals
Figure 4.9
ANciwoikliigliway.
Figure 4.10
Eihernei flame itiagtain.
Network Frames
Each cabling system (or transmission medium) in a packet-
switched network has its own method of conveying or framing
information. These network frames act as “delivery trucks” de-
signed for a specific kind of network highway (see figure 4.9).
“APPLE TALK”
6096
W« Sflif Configure
cr~i
rmft«Mrr
So You Don’t Have To^
zom=m==^
Ofkivcitv
“TPC/IP”
0600
t— 1
Fbnytaile U»
Z_ s
rruvuNffr
czn
rratiuuT
PECnet
6003
Sinco 1976
Novell
6136
Someone Had..
To Po It
With Ethernet, a network frame is composed of a number of
different components, each with its own function. The first two
components are the source and destination of the Ethernet frame
(see figure 4.10). These are identified with the unique 48-bit
Ethernet hardware addresses. They’re similar to the “From” and
“To” logical addresses of an AppleTalk DDP datagram.
Source Destination Type Data CRC
Chapter Four Networking Concepts
The next component is known as the type code. The type code
uniquely identifies the networking protocol contained within the
next component of the frame. Every Ethernet networking protocol
has a unique type code. AppleTalk has several type codes, one for
plain AppleTalk (809B) and one for AARP (80F3). A DECnet packet
has a type code of 6003. Think of the type code as the sign on the
side of the truck that identifies the contents held within.
The actual payload is placed within the data section of the Ethernet
frame. The data are usually protocol specific. In the case of
AppleTalk, the data section is the place where the DDP datagrams
are stored. The data section can vary from anywhere between 5
and 1500 bytes in length.
The last component is used for error checking. A special algorithm
known as a cyclic redundancy check is used to ensure that no
transmission errors have corrupted the data.
NOTE: Because most networking frames can support multiple protocols, it’s
important to remember that networks, such as Ethernet and Token Ring, were
designed to support multiple protocols. Therefore, it’s not mandatory to
restrict your network to a single protocol.
Network Signaling
When binary signals are finally sent to the Cabling layer for trans-
mission, the software that generates the binary communications
data must now interact with the “real,” physical world. While a
complete explanation of the physical laws and the electronics
involved with these network communications is beyond the scope
of this book, an introduction to the basic concepts will help you
understand the capabilities and limitations of the network.
o
Part One Networking Fundamentals
A good starting point is the concept of network speed. Ethernet is
not “faster” than LocalTalk. All electrical signals travel at speeds
close to the speed of light (approximately 80% of it). In order to
transmit digital, or any kind of data, with electromagnetic signals,
there are a limited number of choices available.
One way is to vary the amplitude, or strength, of the signal as time
progresses. By changing (modulating) the amplitude, the ones and
zeros of the binary data can be represented by different voltage
levels (see figure 4.1 1). Actually, amplitude modulation was the
basis for the very first radio signals; AM radio still works this way
today. With AM radio, though, the source of the transmission is not
digital, but tmalog.
Figure 4.11
Willi aiiipliiuile
modulation, itie
amoliiode.oistrengllijf
ilie signal varies Willi
lime.Iheiiequency
remains cnnsiant
<u
■o
Frequency
Ai
/TllAI/k
Time
Another technique is to vary tlie frequency of the signal with time.
Instead of changing the amplitude or voltage, ones and zeros are
represented by shifts in frequency (see figure 4.12). This technique,
known as Frequency Modulation, is a much more efficient and
reliable way to transmit data. Just compare the broadcast quality of
AM and FM radio stations and you’ll begin to get the idea.
In fact, radio and television stations are a good way to describe the
signaling behind computer networking. The network cable, be it
LocalTalk, Ethernet, or Token Ring, has many similarities to a radio
or TV broadcast.
Chapter Four Networking Concepts
Figure 4.12
Wilhiiequency
modiilailDniheampliiiiile
olllie signal mains
conslaniJIieliequency
(ilie distance beiweenihe
peaks and valleys) vaiies
will) lime to encndeilie
signal.
With radio and television, each station broadcasts over a specific
frequency. All these different frequencies are being used simulta-
neously to carry the hundreds of programs that you can watch or
hear. All that’s required is to tune in to the appropriate frequency
with the appropriate receiver.
In computer networking, this kind of multi-channel access is often
referred to as broadband networking. Broadband networks are
common in campus environments and large corporate environ-
ments where multiple communication channels are required (see
figure 4.13).
Channel A
Channel B
Channel C
Channel D
Figure 4.13
Bioadband
cpmmppicaiiopuses
piuliiple channels pnope
signalling medinm.
Most computer networks, particularly local-area networks, use a
single channel technique known as baseband networking (.see
figure 4.14). To compare it to radio and television broadcasting, it’s
as if each station were communicating over the same frequency or
channel.
Part One Networking Fundamentals
Daselianilcoiiiiiiunicaiioii
usesonechaanelto
Ifansmiiinlormaiion.
Figure 4.14
Of course, unless there were strict controls on when each radio and
TV station were permitted to transmit, the single shared channel
would be a garbled mess of colliding and interfering transmissions.
Such is the problem confronted by the designers of modern
network cabling systems — how best to share the common carrier
and maximize throughput. Ethernet and Token Ring exemplify two
of tlie more popular approaches to solve this problem.
With Ethernet, the solution is to play a statistical game of listen and
wait. Wlien an Ethernet node has to send a message, it first “lis-
tens” to the network. If another station is transmitting a message, it
waits until another check indicates that the channel is clear. If the
channel is clear, tlie Ethernet station begins to transmit. The major
problem with this technique is one of timing. Station A can start to
transmit, but its signal might not reach Station B before Station B
begins to transmit. Even though the signal travels at close to the
speed of light, the delay is often enough to fool a station into
transmitting when anotlier message has already started to make its
way onto the network (see figure 4.15).
When this happens on an Ethernet network, it’s called a collision.
Fortunately, the Ethernet hardware can handle and recover from
these collisions. When they occur, the messages are retransmitted
after a short waiting period.
One of the side effects of this collision detection technique is that
the actual throughput of heavily loaded Ethernet networks will
begin to degrade as the number of collisions and retransmits
begins to increase.
Chapter Four Networking Concepts
Figure 4.1 5
Eihernei collisions.
To compare Ethernet networking to our radio and TV example,
imagine that all the radio and TV stations share a common fre-
quency or channel. Before transmitting, each station turns on a
radio or TV to check if anyone else is transmitting. If the channel is
unused, the station begins to transmit for a specific period of time.
Occasionally, two stations might start to transmit at roughly the
same time, and the airw^aves become garbled. But shortly after you
hear a burst of static or the TV screen goes fuzzy, special circuitry
kicks in, transmits the first signal, and stores the second signal for
subsequent playback (see figure 4.16). Of course, as more and
more stations are added, the static bursts and interruptions
become more frequent, until it reaches a point where there's
simply too much traffic — the listeners are getting more static than
programming.
With Token Ring, the technique is a bit different. Each station can
transmit for a specific period of time. The time period is deter-
mined by a token that gets sent around to each of the stations.
o
Part One Networking Fundamentals
When a station possesses the token, then — and only then — is it
allowed to transmit. It’s kind of like the hall pass in Junior High
School that establishes shared use of the rest rooms.
Figure 4.16
As ilie level olirallic
incieasesonanlilieinei
network, a point is
teaclteil where actoal
throoghpotisoptiinireil.
Eventoally, as activity
increases. throughput
Unlike Ethernet, it’s not necessary to "listen” to the network before
transmitting. A welcome side benefit is that the network perfor-
mance doesn’t degrade when it’s pushed to the limit. With Token
Ring, as more and more stations are added to the network, a point
of equilibrium is reached where the network will continue to
deliver its maximum throughput although each station may have
to wait a correspondingly longer period of time before transmit-
ting.
With our radio and TV analogy. Token Ring networking is made
very simple. Each station is allowed to transmit only at specific
times and only for so long. Every station transmits in turn and
every station is assured of access to airwaves (see figure 4.17). As
new stations are added, the time between a given station’s trans-
mission sessions may increase, but at least the listeners won’t be
bothered with interruptions and static.
LocalTalk uses a “wait and listen" approach similar to that of
Ethernet. With Ethernet, this access metliod is called CSMA/ CD. It
stands for Carrier Sense (check for carrier before transmission)
Chapter Four Networking Concepts
Multiple Access (a number of nodes can use the cable at one time)
Collision Detection (there’s special circuitry built into the Ethernet
controller that detects and recovers from collisions).
Figure 4.17
As the level oliiallic
incieasesonaloken
mngnetwoik.ilie
ihroughpui leaches a
siahle point and lemains
cnnstani. Unlike liheinei
nodepiedaiionisseen.
In order to get the cost down, Apple omitted the extra components
required to detect and recover from collisions. Instead, LocalTalk
tries its best to avoid collisions by waiting random periods of time.
If any garbling occurs, the signal is retransmitted by the higher-
level AppleTalk protocols. This technique is called CSMA/CA,
where the CA stands for Collision Avoidance.
Conclusion
The fundamentals of networking can be described by the four
basic layers, but certain aspects of networking require additional
explanation. These technical aspects can also be easUy explained
using convenient and common metaphors.
The basis for most computer networking is the technique of packet
switching. Discrete packets of data are sent over the networking
medium from the sender to the recipient. These devices must be
identified, and that is the reason for a network addressing scheme.
Part One Networking Fundamentals
Depending on the physics of the cabling system, different elec-
tronic access methods must be used. Ethernet is an example of a
delivery mechanism that detects for the presence of a signal,
listens for traffic, and then begins to transmit. Other cabling
systems, such as Token Ring, parse the available bandwidth by
specifying a specific period for transmission.
o
Common
Network
Components
he networking industry is full of terms and
buzzw'ords that are confusing and often mislead-
ing. Using NetPlCTs, this chapter unambigu-
ously defines and depicts common Macintosh
networking components such as bridges, repeat-
ers, and gateways. Once these basic components
are defined, they will be placed in the context of
various netw'orking topologies. These various topologies
and networking implementations, such as Local Area Networks
and Wide Area Netw'orks, will be discussed in Chapter 5.
LANs & WANs
LAN and WAN are acronyms, respectively standing for Local Area
Network and Wide Area Network. Distance is the major difference
between the two. LANs are usually confined to a single building or
a group of buildings in the same general vicinity. When multiple
buildings are involved, it’s common to have a LAN network seg-
Part One Networking Fundamentals
ment in each building. These segments are often connected with
other segments. Underground fiber optic cable is frequently used
to connect the segments.
WANs are interconnected networks that span wide geographic
areas. Basically, this means distances greater than what can be
supported by a LAN. While there are no definite rules, WAN
distances are generally measured in miles and LAN distances are
usually measured in feet.
Tj'pically, W'AN connections have a limited bandwidth when
compared to IAN connections. This limitation might be one or two
orders of magnitude. For example, an Ethernet network has a
maximum capacity of 10 million bits per second, or 10 Mbps. Tl, a
popular WAN carrier, has a bandwidth of 1.544 Mbps. In this case,
the Ethernet network has roughly 6 1 12 times the bandwidth of the
Tl link.
WAN links, such as Tl, can be very expensive. These connections
are usually leased by phone companies or other service providers
and may cost several thousand dollars per month to maintain.
Clearly, optimization and rational use of these WAN resources is
important. Later on, we'll discuss the specific issues and problems
with AppleTalk protocols on wide-area networks.
Repeaters are the simplest of networking devices. They exist at the
very bottom of the stack, functioning at the Cabling layer — the
Physical layer of the OSl Reference Model (see figure 5.1). At this
layer, the repeater is entirely ignorant of the higher-level protocols,
such as AppleTalk. Repeaters only deal with the electrical signals
that exist at the Cabling layer. As such, repeaters are often referred
to as being protocol transparent .
Chapter Five Common Network Components
LocalTalk Repeater
Figure 5.1
KeinCIolalocaM
repeaiet.
Repeaters are essentially signal boosters that amplify weakened
incoming signals, thus extending the effective range of a cable.
Since repeaters are "dumb” devices, they also will amplify noise or
any unwanted signals that might find their way onto your network.
LocalTalk Repeaters
There are a number of LocalTalk repeaters on the market. Some of
them, such as Farallon’s PhoneNET Repeater, simply extend single
segments of PhoneNET/LocalTalk. Other repeaters (for example,
Farallon’s StarController) have multiple ports and permit different
wiring topologies.
The simplest kind of LocalTalk network is when all the participat-
ing devices are connected in a daisy-chain fashion. However,
daisy-chaining can create problems for all but the smallest
LocalTalk networks.
With daisy-chained networks, the wiring runs from node to node,
creating the real possibility of occasional disconnects and disrup-
tions to the network. By using a star or radial topology with a
multiport LocalTalk repeater, each branch of the star is indepen-
dent from the others (see figure 5.2).
0
Part One Networking Fundamentals
Figure 5.2
Radial repealer topology.
Punch-Down Blocl<
LocalTalk
Repeater
Think of repeaters as wiring conveniences. They do not filter or
limit nem'ork traffic in any way. The same signal and network
traffic is seen on all ports of the device.
It is possible to construct very large LocalTalk networks by inter-
connecting repeaters, but problems often occur as the available
bandwidth is still being shared across all wiring segments. At some
point, the limited bandwidth of LocalTalk simply cannot support
the traffic that’s routinely generated by an AppleTalk network. So,
it’s best to limit the use of LocalTalk repeaters to extending that
extra long run for the people in the shipping department, and to
break away from the nasty habit of daisy-chaining.
Ethernet Repeaters
Like their LocalTalk cousins, Ethernet repeaters are protocol-
transparent wiring conveniences. Like the multiport LocalTalk
repeater (such as Farallon's StarController and Focus’s TurboStar),
Ethernet repeaters let you avoid the daisy-chain network. There
are Ethernet repeaters for all flavors of Ethernet cabling: thick,
thin, and twisted-pair (see figure 5.3).
In fact, with twisted-pair Ethernet (othei*wise known as lOBase-T)
repeaters or hubs are mandatory. All lOBase-T connections are
made through a hub.
Chapter Five Common Network Components
Figure 5.3
Eilierneliepeatef.
Bridges
Bridges function at the upper Cabling layer, or the Data Link layer
of the OSI Model. They are one step higher tlian repeaters on the
network evolutionary scale.
Bridges use the layer that connects a specific networking protocol
to a specific cabling system. Because they function one layer
higher than repeaters, and thus know about physical hardware
addresses or cable-specific addressing, bridges are often able to
use these addresses to optimize traffic (see figure 5.4).
LocalTalk Bridge
Figure 5.4
Bridge NelPICI.
Bridges can examine the source and destination of netv\^ork frames
and maintain a listing of which nodes are on which port. Then, the
0
Part One Networking Fundamentals
bridge can minimize traffic by only passing those frames that are
destined for nodes on the other side (see figure 5.5).
Hgure 5.5
Ettierfleibfidoes.
08 - 12 - 32 - 19-32 08 - 12 - 32 - 19-33 08 - 12 - 32 - 19-36
Figure 5.6
Etonelbridoelilleiino.
Like repeaters, bridges are often used to extend a particular type of
network. Unlike repeaters, some bridges also can use the cabling-
specific information to filter packets based on their protocol type.
Certain Ethernet bridges are capable of examining the Ethernet
Type Code of a frame. These bridges also can be programmed to
pass or block a network frame based on the Type Code. These
bridges are often referred to as filtering bridges (see figure 5.6).
08 - 12 - 32 - 19-32 08 - 12 - 32 - 19-33 08 - 12 - 32 - 19-36
o
Chapter Five Common Network Components
Remember, bridges are cable-specific and, like repeaters, are
essentially protocol-transparent. Bridges don’t inspect the
datagram to see the logical address data. They can only infer, as in
the case of Ethernet Type Codes, the networking protocol con-
tained within.
LocalTalk Bridges
At this time, there is only one LocalTalk-specific bridge. Tribe's
LocalSwitch uses the LocalTalk-specific node number to keep track
of node locations. It then uses these numbers to limit traffic to only
those segments containing the valid nodes.
For example, if segment A of the Tribe bridge has LocalTalk nodes
1 and 2, and segment B has nodes 3 and 4, then whenever nodes
1 and 2 communicate, the resultant traffic will only be seen on
segment A and not on segment B.
Ethernet Bridges
Historically, Ethernet bridges have been used to interconnect large
corporations and organizations over long distances. By using
bridges that connect to Ethernet on one side and wide-area
network services on the other, organizations easily could connect
two geographically remote Ediernet LANs and make them appear
as a single connected entity (see figure 5.7).
Unless some form of packet or frame filtering is employed, these
devices logically connect the segments. Certain types of traffic seen
on one segment will eventually be seen on the other segment. Of
course, if the traffic represents frames destined for the other side,
this makes sense. All too often, however, the traffic seen on one
side of the bridge is local and is not destined for the other side. This
unwanted passage of traffic particularly creates problems when the
interconnection between the two networks is at a restricted
bandwidth. This often creates a bottleneck between the higher
bandwidth of the LANs.
o
Part One Networking Fundamentals
Rgure 5.7
OneiiiDlUieinet
This network
is conceptually
the same...
...as this
network...
...particularly when it comes to network broadcasts.
Figure 5.8
WANbotlleneck.
For example, it is a common practice to connect two Ethernet
LANs with a leased telephone line, with a bandwidth of 56 Kbps.
Considering that Ethernet has a bandwidth of 10 Mbps, it’s like
having a 179-lane freeway being reduced to a single lane of traffic
(see figure 5.8).
Therefore, if you're using wide-area bridges over modest band-
width connections, it’s very important to keep unwanted or
unneeded traffic to a minimum. This shortcoming, inherent with
wide-area network bridges, is one reason why the wide-area router
is fast replacing the bridge as the wide-area network device of
choice.
o
Chapter Fine Common Network Components
Routers & Brouters
Routers do their work at the Protocol layer, inspecting the source
and the destination of a specific protocol’s datagram. Routers, like
the children passing notes in the classroom, make decisions on
how best to route messages based on the source and destination
address. This is fundamentally different from bridges, which
process network frames without regard to the networking protocol
contained within.
Because routers operate at the Protocol layer, they are often
referred to as being protocol-specific. There are AppleTalk routers,
DECnet routers, TCP/IP routers, and Novell routers. There are
even routers that can route multiple protocols concurrently.
Each connection on a router is called a port (see figure 5.9). These
ports may be the same cabling type, or they may be different. For
example, there are many AppleTalk routers that have a LxjcalTalk
port and an Ethernet port.
Hgure 5.9
PoiisoiaiQuiei.
Because of this, it is often assumed that these devices are simply
used to connect a LocalTalk network to an Ethernet network. In
addition to providing this basic cable conversion, as shown in
o
Part One Networking Fundamentals
figure 5. 10, routers also possess a special capability. Routers can be
used to isolate traffic.
Figure 5.10
Houietuonneciing
muliiple cable lypes.
Similar to the filtering bridges mentioned before, routers are able
to keep local traffic local. For example, if a Macintosh and a
LaserWriter are connected to a network on port A, the network
traffic generated between these two devices will be limited to that
network (see figure 5.1 1). The router will forward a datagram to
another network only if the destination of the datagram is on that
network.
So, routers perform two basic functions. They provide a funda-
mental way to connect one network to another, and they effec-
tively isolate nenvork traffic.
o
Chapter Five Common Network Components
Figure 5.1 1
Network itallic isolation
witi) routers.
AppleTalk Routers
What started in 1985 with the Kinetics FastPath has evolved into a
crowded and competitive market with dozens of AppleTalk rout-
ers. While these routers support different cabling choices, unique
configuration software, and value-added options, every AppleTalk
router works on the same basic principles. All AppleTalk routers
use AppleTalk network numbers to differentiate their ports. In fact,
for most AppleTalk networks, the network numbers originate in
the router.
Let’s use a simple example of two LocalTalk networks connected
by a two-port LocalTalk-to-LocalTalk AppleTalk router. In this
example, we’ve used the configuration software that came with the
router and established unique network numbers for each of the
ports. It’s crucial that each network number be unique. This can be
a challenge for a very large network. With large networks, it’s
necessary to plan and administer AppleTalk network numbers. If
an AppleTalk network number is duplicated anywhere on the
collective network, unpredictable results may occur.
o
Part One Networking Fundamentals
AppleTalk routers can be connected in several different topologies.
They can be arranged in a serial fashion, where one network is
simply connected to the other. In Figure 5.12, network number 2 is
serially connected to 1 and 3.
Figure 5.12
Ihieeseiiallyconnecleil
AppleTalk neiwoiks.
Therefore, in order for an AppleTalk datagram to travel from
network 1 to network 3, it must go through two routers. Each time
a datagram passes through a router, it is called a hop. A hop is a
unit of measurement that is used to measure the distance between
the sender and recipient. AppleTalk has a maximum hop count
of 15.
Using serially connected routers to build large AppleTalk networks
can be problematic because every time a datagram travels through
a router, it must inspect the network address, which takes a certain
amount of time. In a large serially-routed network, these delays
accumulate. Another problem with this topology is that networks
that are located between the sender and recipient have to deal with
all the network traffic.
A better way to build large AppleTalk networks is to use a backbone
topology (see figure 5.13). Here, a common backbone network acts
like a common network that connects a number of routers. With
o
Chapter Five Common Network Components
this topology, no network is more than two hops away. Backbone
topologies are very common in large AppleTalk networks, where
the backbone is often an Ethernet network and the routers are
used to connect a number of LocalTalk networks to the backbone.
Network #1
-®-
q t;i q
5
Network #2
Network #3
< 5 ^
Figure 5.13
IhreeApplelalkneiwoiks
OP a backbone.
As the cost of Ethernet connections continue to plummet, and
since Apple now offers computers and printers with built-in
Ethernet connections, LocalTalk networks and routers are becom-
ing less popular. Instead, Ethernet-to-Ethernet routers are begin-
ning to displace the sales of LocalTalk-to-Ethernet routers. These
routers are not used to connect dissimilar cabling; the goal is
simply to isolate network traffic.
For many networks, routers provide a convenient way to isolate
areas of high network traffic. Let’s use the example of an organiza-
tion with an Ethernet backbone. Each department within the
organization has its own Macintosh computers and LaserWriter
printers connected to the Ethernet. The desktop publishing group
is continually taxing the network with demanding print jobs and
the transfer of large image files. The other departments and users
have started to complain about poor network performance. One
way to alleviate the problem is to place the DTP group on a sepa-
rate Ethernet, connected to the primary network with an Ethernet-
to-Ethernet router (see figure 5.14). With this approach, the
o
Part One Networking Fundamentals
extensive traffic associated with the DTP group will be kept to its
own network.
DTP Group
The network numbers maintained by the routers are the primary
determinants of an AppleTalk network. As mentioned before, the
network numbers are used by the AppleTalk software to assist with
the routing of datagrams. The problem with network numbers is
that they’re not the best way for users to locate network services.
To help users find services, Apple devised a way to associate names
with networks. These names are called zones. AppleTalk zone
names are stored in the router along with the network numbers.
Unlike AppleTalk network numbers, which must be unique,
AppleTalk zone names can be duplicated throughout the network.
This feature makes it possible to organize network services in
logical groups.
As an example, imagine a ten-story building with a backbone
Ethernet running up through the building. In the wiring closet of
o
Chapter Five Common Network Components
each floor, a LocalTalk-to-Ethernet AppleTalk router is used to
connect a number of Macs and LaserWriters to the backbone. Each
LocalTalk network has a unique network number and a zone
name. The LocalTalk network on the first floor is network number
1 , and the zone name is “ 1st Floor.”
The Ethernet backbone also carries AppleTalk traffic, and therefore
is also considered an AppleTalk network. In this example, the
backbone is assigned network number 1000 and given the zone
name of “Backbone.” This is accomplished by configuring the
routers so that all routers agree that the backbone is defined as
network 1000 and has the zone name of “Backbone.”
Most floors only have a dozen or so devices, so the theoretical limit
of 32 LocalTalk devices per network does not matter. But on the
tenth floor, there are more than 50 devices that need to be con-
nected to the network. Since this is too many devices to be placed
on a single segment of LocalTalk, another LocalTalk segment and
router will be required.
Like the other routers, this additional router on the tenth floor
must have a unique network number. Since it will be convenient
for the users to refer to this new network with the same zone name
as the other network, the zone name also will be “10th Floor” (see
figure 5.15).
What’s interesting about this process of zone grouping is that the
AppleTalk protocols will combine all like devices by zone. So, if
there were five LaserWriters in each of the two networks on the
tenth floor, the Chooser would combine the devices and display a
list of ten printers for the “10th Floor” zone.
o
Part One Networking Fundamentals
Rgure 5.15
Network #100 Zone: "10th Floor
^ ^ ^ ^
Network #10 Zone: "10th Floor'
® c
^ o
^ «
oP
5'.
Network #2 Zone: "2nd Floor"
■< 5 >
(D
O
C
o
N
Network #1 Zone: "1st Floor’
<E>\
Multiprotocol Routers & Brouters
Certain routers are capable of routing multiple networking proto-
cols at the same time. These are generally known as multiprotocol
routers. A good example of such a device is the Shiva FastPath.
The FastPath is a LocalTalk-to-Ethernet AppleTalk router, but in
addition to AppleTalk, it can also route DECnet and TCP/IP
protocols. A multiprotocol router does not convert one protocol
into another: it simply routes each protocol that comes along
according to the protocol type.
When the FastPath encounters DECnet datagrams, it routes them
in accordance to the DECnet routing rules and configuration data
maintained in the router’s program and memory. When an
AppleTalk datagram appears, it’s routed using the AppleTalk rules,
It’s as if three separate routers — one each for AppleTalk, DECnet
Chapter Five Common Network Components
and TCP/IP — were merged together as a single unit, sharing their
hardware and portions of their configuration software.
With a multiprotocol router, each protocol must be configured
separately. So, in order to set up a FastPath to route all three
protocols, you must configure each protocol independently using
the FastPath Manager application that comes with the router.
While AppleTalk, DECnet and TCP/IP are all different protocols,
they do have something in common. They all can be routed. This is
because all three protocols utilize the concept that large networks
can be broken down into smaller pieces.
With AppleTalk, large networks, called internetworks, are subdi-
vided into networks by network numbers. The AppleTalk protocols
support more than 65,000 networks, each potentially supporting
either 253 or 254 nodes (under Phase 2 or Phase 1 AppleTalk,
respectively).
DECnet uses a similar technique where large networks are broken
down into areas. DECnet supports a maximum of 64 areas; each
area can potentially support 1024 nodes. TCP/IP, like AppleTalk
and DECnet, also breaks down large networks into smaller group-
ings (see figure 5.16).
Figure 5.16
Muliiproiocol routers.
Part One Networking Fundamentals
Each of these protocols utilize numbers that will be used by their
respective routers to subdivide large networks. Each subdivision
enjoys the benefits of traffic isolation, so that unneeded and
undesirable traffic is kept within each subnetwork.
Returning to our analog^' of students passing notes in class, imag-
ine that tlie students form three cliques, and members of each
clique never talk to the other cliques. To accommodate the note
passing, the students decide that each clique will use a different
color of paper for their notes. Also, each clique develops certain
rules and idiosyncrasies for passing its notes. Those students who
are allowed to pass all three colors of notes know the rules for each
clique's notes. For example, the red notes have the last name first
and have to be passed with the palm held down; the blue notes are
addressed with just initials and have to be transferred within a
closed fist.
The students that know all the rules act as multiprotocol routers.
Each clique has tlieir own protocol with addressing differences
that have to be managed by the routing student. He has to be able
to deal with notes that use different protocols in order to send
them along to their proper destination.
Some protocols don't use routing numbers at all. These protocols
rely on a single monolithic network, with each node having a
unique identifier. One such protocol is DEC'S Local Area Trans-
port, or LAT. LAT is primarily used to provide terminal services to
networked devices, such as ‘"dumb” VT terminals, connected to an
Ethernet network with a device known as a terminal server. The
terminal server takes the asynchronous communication from the
VT terminal and places it onto the Ethernet within a LAT frame.
The terminal server is a computer that can direct the LAT traffic to
any available host (usually a DEC VAX) on the Ethernet. PCs and
Macs equipped with an Ethernet card can also speak LAT directly
to networked hosts.
Chapter Five Common Network Components
Unlike AppleTalk, DECnet, and TCP/IP, however, LAT was de-
signed to run over a single, logical segment of Ethernet. There is
nothing equivalent to a LAT network number or area number.
Therefore, LAT is an example of a non-routable protocol. There is
no such device as a LAT router — there is no network or area
number that the router could use to route datagrams. Since LAT
cannot be routed, it is often difficult to expand LAT terminal
services beyond the confines of a single Ethernet LAN.
One way that a LAT network can be expanded to another LAN is
with a bridge. Since bridges only look at the Cabling layer — and do
not care about the protocol-specific datagram information — they
can be used to extend an Ethernet and to bridge LAT ser\dces
between remote Ethernet segments. As mentioned before,
Ethernet bridges can be used to essentially create one large
Ethernet.
What if we wanted to design a network providing AppleTalk,
DECnet, TCP/IP and LAT services to two sites: one in New York
City and the other in San Francisco (as in figure 5.17)? Each site has
a backbone Ethernet.
Figure 5.17
NY loSFO with bridged
Eihernei.
SFO
Part One Networking Fundamentals
If we simply connect these two networks using Ethernet bridges
and leased telephone lines, everything would work, but we would
have essentially created one big Ethernet. All the protocols will
freely traverse both networks via the phone lines. This might create
a problem. Every time someone prints a document or sends a mail
message in New York City, the traffic generated by those activities
is sent through the phone lines and reproduced on the San Fran-
cisco network. Filtering bridges would help somewhat, by keeping
track of what Ethernet devices are on what side of the connection,
but certain kinds of traffic (known as Ethernet broadcasts or
multicasts) would still be seen on both sides of the network.
To really solve the problem, a pair of multiprotocol routers could
be used to subdivide our AppleTalk, DECnet, and TCP/IP networks
into logical segments on each side of the connection. This would
effectively isolate any unwanted traffic, preventing it from travers-
ing the expensive and restrictive leased phone lines.
If we simply added the routers in addition to the Ethernet bridges,
we would have a problem (see figure 5.18). The AppleTalk, DECnet,
and TCP/IP traffic would still be able to sneak through the bridge
and get to the other side. We simply can’t do away with the bridge
because that’s how LAT is getting to the other side. And since LAT
is a non-routable protocol, it can’t be routed through the multi-
protocol router.
One solution would be to configure each bridge to filter out and
ignore the AppleTalk, DECnet, and TCP/IP traffic (see figure 5.19).
This would essentially limit the bridge to only carrying LAT traffic,
and the router would then handle the AppleTalk, DECnet, and
TCP/IP traffic.
Chapter Five Common Network Components
fl-
AppleTalk, DECnet, TCP/IP
lB
Q-
. ■ —
1 in —
i HMRnti y - ■ ■ ^ 1 eo»wwrt 1
s
CE-
1 erldg* 1 1 irtdp. |
AppleTalk, DECnet, TCP/IP
and LAT
nn'
SFO NYC
ngure 5.18
Tlierouiei is passing
ApplelalUECneianil
ICP/IP,butsoisilie
bridge.
Figure 5.19
ApplelalUECneiaad
ICP/IP are filtered at ibe
bridges.
Router manufacturers, in an attempt to solve this problem, have
designed hybrid routers and bridges. These devices, sometimes
referred to as brouters, route all the routable protocols and bridge
the non-routable protocols. Most of the high-end routers (such as
Cisco and Wellfleet) can be set to act as multiprotocol brouters.
Another approach to solve this problem is to place the non-
routable protocol inside another protocol that can be routed. This
general practice is called tunneling, and can be used to solve a
Part One Networking Fundamentals
number of different problems in addition to LAT. AppleTalk
packets can be tunneled inside DECnet and TCP/IP packets for
several reasons. This practice will be discussed later in the book in
more detail.
Figure 5.20
Gateway Iheofv.
Gateways
Gateways are the most complex of network devices, functioning at
the upper-most layers where the format of the data comes into
play (see figure 5.20). Simply put, gateways provide translation
services on the network. They can be used to connect different
kinds of network cabling, but most important is that gateways
convert the format in addition to the networking protocol. Gate-
ways are the language translators of the network.
Network
Drivers
A O
Common or Different Physical Media
Different Networking Protocols
(i.e. AppleTalk, DECnet, TCP/IP)
O □ A
Since gateways build upon the bottom two layers, it’s common
that most gatew'ays also function as multiprotocol routers. Some
Computer A Gateway Computer B
Programs
_7V
A
LA.
O
n]
o
Programs
n.
o
Chapter Rue Common Network Components
gateways can even be exclusively set up as routers, thus ignoring
their higher-level conversion capabilities.
The main advantage and purpose of gateways is to provide a
centralized alternative to placing a “foreign" transport and service
on your client computers.
AppleTalk Gateways
Suppose you needed to connect a Macintosh to a UNIX-based
workstation, perhaps to share files. One approach would be to
simply equip your Macintosh with the TCP/IP protocol stack, and
then choose a file service that works in concert with TCP/IP, such
as NFS (Network Filing System).
Once equipped, the Macintosh would match the UNDC workstation
at all layers, and therefore would be able to transfer files. Of course,
the other solution would be to equip the UNIX machine with
AppleTalk and its file sharing software, AFP (Apple Filing Protocol).
But, if we wanted to stick with TCP/IP and NFS on the UNIX
machine, we would still have another option. With an AFP-
AppleTalk/NFS-TCP/IP gateway, the Macintosh would use the
gateway to convert AFP and AppleTalk protocols into correspond-
ing NFS and TCP/IP protocols.
To the Macintosh, the AFP-AppleTalk side of the gateway looks like
any AFP or AppleShare file server. To the UNIX workstation, the
NFS-TCP/IP side of the gateway looks like an NFS client. The
gateway handles the complete conversion between the different
filing systems and network protocols (see figure 5.21).
o
Part One Networking Fundamentals
Figure 5.21
KeiPICIol NFS gateway
(Cayman Gaioilioxl.
Macintosh
UNIX Workstation (NeXT)
Figure 5.22
NeiPICIoiAgplelalk/
DECnei gateway.
One appeal of a gateway is that no special software is required on
either the client or tlie server. The gateway performs all necessary
functions. This approach may not be as fast as placing the same
native protocols on each machine, but it can be a simple and cost
effective way to share dissimilar network services. There are several
different kinds of gateways for the Macintosh. In addition to the
AFP/NFS gateways, which are available from several vendors, there
are AppleTalk gateways that work other protocols as well.
In dec’s PATHWORKS for Macintosh product, there are two such
gateways. One gateway is tlie AppleTalk-DECnet gateway. This
gateway runs on the VAX, in conjunction with AppleTalk for VMS
and its native DECnet. Macintosh services, such as DEC’s All-In-1
Mail for Macintosh, can use the AppleTalk protocol to communi-
cate with the VAX. However, DEC’s mail server doesn’t communi-
cate over the AppleTalk for VMS protocol stack — it uses DECnet. So
the gateway converts the protocol and service calls from AppleTalk
to the corresponding DECnet calls (see figure 5.22).
Macintosh DEC VAX
The other PATHWORKS for Mac gateway is the AppleTalk LAT
Gateway (see figure 5.23). It enables LocalTalk-equipped Macs
Chapter Five Common Network Components
(or any non-Ethernet resident Macs, such as dial-in AppleTalk
Remote Access Macs) to access LAT-based terminal services. This
gateway runs in the background of an Ethernet-resident Macin-
tosh. Client Macs use the AppleTalk- LAT tool to connect to the
gateway; the gateway takes the incoming AppleTalk data, converts
it to LAT, and sends it out over tlie Ethernet.
Macintosh
MacTermlnal 3.0
Apple Macintosh
AppleTalk / LAT Gateway
VAX Application
Formats
) / Formats V-
( Formats ) (
Formats
AppleTalk
1 1 AppleTalk ] I
f 1 1
LAT
» LocalTalk
/ S LocalTalk /
S Ethernet / ^
► Ethernet <
Figure 5.23
Applelalk/lAI gateway.
In all these examples, one protocol has been converted and trans-
formed into another. To help with the basic idea behind gateways,
let’s return to the analogy of students passing notes in a classroom.
Before, we discussed how packet-switched networks could be
likened to students passing notes in a classroom. When a student
looks at the “To” and “From” on the note and passes the note
along, she is essentially performing the function of a router. But
let’s imagine a situation where one of the students, Sonia, can only
understand Polish and does not understand English.
For Sonia to communicate with the other students, she can learn
English, or the other students can learn Polish. Until this happens,
however, Sonia must rely on a gateway. Her sister, Patrizia, under-
stands both Polish and English. Notes that are sent to Sonia must
first be sent to Patrizia, who converts the English to Polish, and
then passes the note to Sonia. When Sonia sends a note to one of
the other students, Patrizia reverses the process and converts the
Polish into English.
In this brief example, we can see the difference between a router
and a gateway. Those students who merely passed the notes along
Part One Networking Fundamentals
acted as routers. Patrizia, who converted from one format to
another, acted as a gateway for Sonia.
Miscellaneous: Hubs, Concentrators,
and So Forth
Hubs and concentrators are wiring conveniences that essentially
perform the same function — they provide a central attachment
point for network devices. Unless they Incorporate bridges or
routers, they do not interact with the networking addressing or
protocols in any way. Most large network installations employ
some sort of network concentrators in order to implement a radial
or star wiring topology. The specifics of the various networking
topologies will be covered in the next section.
Usually, network hubs and concentrators are installed in a stan-
dard 19-inch rack. These racks are standardized metal frames that
hold electronic equipment and permit easy access and mainte-
nance. Often, other networking components (such as modems,
bridges, routers, and gateways) are also stored in the rack with the
concentrator. This makes it easy to interconnect the LAN with the
desired devices.
Many hubs and concentrators come wdth diagnostic software that
assists with the management of network devices and performs
fault diagnosis. This software should not be confused with the
protocol-specific configuration software that comes with routers
and gateways. Instead, this software is often able to turn on and off
the concentrator's ports, perform line quality checks, and even
perform rudimentary traffic analysis.
Chapter Five Common Network Components
Network Topologies
One topic that’s covered in all nettvorking books is topology. The
topology of a netAvork is simply the form and sequence of connec-
tions. The topology does not depend on the kind of wiring or
protocols used. There are several different kinds of networking
topologies, each with their own benefits and disadvantages, and
these can be combined to create hybrid networks.
Daisy Chain
One of the simplest network topologies is the daisy chain. As the
name implies, a daisy-chain network is created by simply connect-
ing one node to the next. Perhaps the best example of a daisy-
chain network is LocalTalk.
With LocalTalk cabling, the connections are made with segments
of wire, in a serial fashion — one after the other (see figure 5.24).
Daisy-chained networks are fine for small workgroups where the
wires can be conveniently routed. The problem with daisy chaining
is that the entire network can be easily severed and disrupted if the
cable is broken or disconnected at any point along its length.
Figure 5.24
Oaisy-thainiopologv.
wiililocallalk.
This often happens when someone disconnects their Mac from the
LocalTalk segment by disconnecting either of the two network
segments instead of the single attachment point at the printer port.
It also happens when a new node is added to the network and
cannot be added to one of the two free ends of the network.
o
Part One Networking FundamentaJs
Figure 5.25
Daisy-cliain,lliiii«iie
lopolooy.
Thinwire (or lOBase-2) Ethernet networks can also be set up in a
daisy-chain topology, but the same problems exist (see figure 5.25).
The network can be easily severed by a careless disconnection. The
end points of a thinwire Ethernet network must be terminated with
resistive end caps. If the network is broken at either of the two
network connections on the tee, the resistive load is missing and
the network is disrupted.
won't disrupt the network... will disrupt the network.
Apple recently introduced a version of thinwire Ethernet, where
the resistive loads are automatically maintained in the event of a
disconnection. (At least this way you’ll still have two functioning
network segments.)
For most office and business settings, daisy-chained networks are
usually avoided because of the problems associated with frequent
disconnects and maintenance of the wiring.
Common Bus or Backbone
A common bus, or backbone, network is simiiar to the daisy-chain
network except that the cable connecting all the participating
devices is one entire segment. The nodes connect to the backbone
with an attachment unit.
Chapter Five Common Network Components
The most common example of a backbone network is a thickwire
or lOBase-5 Ethernet cable. A thickwire Ethernet backbone can be
as long as 500 meters (1600 feet). Nodes and other network devices
attach to the thickwire Ethernet with transceiver devices that
clamp onto the cable.
These transceiver connections are somewhat expensive (ranging
anywhere from $100 to $300) and take some time to install. Be-
cause of this, these transceiver connections are very often used to
connect hubs or other devices where multiple connections can be
made at a single location.
Most large network installations that use Ethernet utilize Ethernet
backbones that are connected to other devices. These backbones
often run up through the stories of the building, between the
wiring closets where the connections are established (see figure
5.26).
Wiring Closet
Wiring Closet
Thinwire (10Base-2)
Repeater
Hgure 5.26
Backbone or bus
loiioloiif.
Radial or Star
A radial, or star, topology has connections that radiate from a
single location. Usually there is a single node on each branch of the
Part One Networking Fundamentals
Figure 5.27
Slariopology.
star. Some configurations of star networks support the daisy-
chaining of devices, but this often defeats the principal advantages
of the star topology.
One advantage of star networks is ease of wiring. Each node (or
office) is wired so that all connections converge at a single point —
usually a concentrator located within a wiring closet. Then, when
wiring changes are needed, they can take place within tlie closet.
This avoids the hassle associated with daisy-chaining. Another
advantage is that connections and disconnections can be easily
made without disrupting the other members of the network. A
node could blow up and it still wouldn’t affect the traffic on the
other segments of the star.
Star networks are common in the Macintosh world. With LocalTalk
netvi'orks, there are several repeaters that permit the use of star
topology (see figure 5.27). Farallon’s StarController and the
TurboStar from Focus are two well-known examples. These
devices also come with software which provides some measure
of network maintenance.
There are also Ethernet devices that permit the use of a star topol-
ogy. There are thinwire Ethernet hubs that can be used to get away
from the problems associated with thinwire daisychaining. The
Chapter Fiue Common Network Components
most popular Ethernet hubs today are the lOBase-T (or Uvisted-
pair) Ethernet hubs. These hubs typically attach to a backbone
Ethernet and provide several (usually eight) connections for
twisted-pair Ethernet nodes.
Twisted-pair Ethernet has a maximum range of 100 meters (320
feet) and uses inexpensive four-conductor telephone-style wiring.
Each node must connect to the hub.
Ring
A ring topology (see Figure 5.28) connects all the devices in a
circular manner. Ring networks are typified by Token Ring and by
FDDI, which uses dual rings for redundancy and increased
throughput.
r
!x
Figure 5.28
Ringiopology.
Composite
Most large networks use a combination of topologies (see figure
5.29). Each topolog>' is used for a specific reason and purpose.
Performance, maintenance, and cost are all factors that determine
the choice of a comprehensive networking topology. For example,
it's common to see star and ring networksthat connected to a
backbone network, that’s in turn connected to another backbone
in a serial fashion.
Part One Networking Fundamentals
Figure 5.29
Composite lopology.
Thinwire
Ethernet
Repeater
Macintosh Running
the Appie Internet
Router between
Ethernet and Token
Ring segments
Ethernet Backbone
Token Ring
Conclusion
Part of the difficulty in understanding networking is the plethora of
networking black boxes that perform unknown or mysterious
tasks. These components fall into a number of categories, such as
repeaters, bridges, and routers. Of these, one of the key devices is
the AppleTalk router. It’s the basis for creating large AppleTalk
networks and isolates traffic to a specific segment. Many AppleTalk
routers are also capable of routing other protocols, such as DECnet
and TCP/IP. These are known as multiprotocol routers.
These routers, along with other devices (such as bridges, repeaters,
and gateways), are connected in specific arrangements referred to
as topologies. These topologies are often dictated by the choice of
a cabling system. Small networks, such as LocalTalk networks, are
comprised of a single topology. Most large networks employ
several topologies, such as a combination of backbone and star
topologies, to meet specific networking and wiring requirements.
lats, Transports, and Media.
“Two”
Pai t ll introduces Macintosh spet
networking by breaking the topic
Part II introduces Macintosh specific
four basic components: Services,
networking by breaking the topic into
Formats, Transports, and Media.
the four basic components: Services,
Formats, Transports, and Media.
Macintosh
ig by breaking the topic into
asic components: Services,
Transports, and Media.
Macintosh
Services
his chapter covers the reason why Macintosh
and AppleTalk networking is so popular: the
services. The only reason to network your
computers is to supply services, such as file and
I print services, to the users. The common
Macintosh and AppleTalk-based services will be
described.
AppleShare (AFP)
To many Macintosh users, access to an AppleTalk network means
shared, controlled access to files. To provide such access, Apple
developed a Service layer (Presentation layer in the OSI model)
protocol called the Apple Filing Protocol, or AFP. The AFP server at
the Serv'ice layer represents an implementation of the AFP proto-
col, which can be described as a Format layer protocol in the
NetPICT diagram, or as a Presentation layer format in the OSI
model. Figure 6.1 shows a NetPICT diagram of an AFP Server. In
this example, the AFP server is running on a Macintosh with an
Ethernet card.
Part Two Macintosh Networking
Figure 6.1
WICIofaMacinlosh
luaning AppleShare.
Macintosh Client AppleShare Server
When Apple introduced AFP, Macintosh users were introduced for
the first time to such radical concepts as user names, passwords,
groups, ownership, and protection. AFP provided a mechanism
whereby Macintosh users could find AFP servers on the network,
log on these servers, and create, edit, and share files. Apple called
this product AppleShare. AppleShare was easy to use, because the
files stored on the server appeared as an extension of the user’s
local hard disk. All the techniques learned in creating and naming
folders and files were immediately transferable to the file server
environment.
AppleShare is a textbook example of a client/server connection. A
user on a client Macintosh identifies and connects to a Macintosh
that is running the AppleShare server software. This is done by
clicking on the AppleShare icon in the Chooser.
AFP Server
Actually, it would be more correct if the icon were labeled AFP Server, because
all AFP servers on all platforms will be searched for— not just the Macs
running AppleShare.
Once the AppleShare icon is highlighted, as shown in figure 6.2, the
Macintosh sends out a series of special inquiry packets that look
for all the AFP servers in the current zone. The AFP servers, hearing
Chapter Six Macintosh Services
this request, respond to the sender with their network address and
name. The names are displayed in the Chooser and the addresses
are remembered for future communications.
Chooser !
Select a file server:
llannaTuva
Henry’s Hideaway
Stella’s Trash Can
AppleTalk
® Active
O Inactive
7.2
Figure 6.2
Itie Macintosh Chooser,
showing AppleShare
servers.
The user then selects the desired server and enters a user name
and password in order to gain entry to the server. Once the user
has successfully logged onto the sei-ver, a list of volumes — shared
disks — appears in a dialog box. Multiple volumes can be created to
address the particular needs of the workgroup. AppleShare comes
with a management utility that is used to add and delete users and
perform other necessary administrative tasks.
System 7 File Sharing
Starting with System 7, Apple expanded the concept of AFP ser-
vices so that any Macintosh can be an AFP server and client. This is
known as Macintosh File Sharing. Assuming your Macintosh is
running System 7, you can grant or deny access to your Macintosh
on a user or group (a collection of users) basis. You then designate
Part Two Macintosh Networking
Figure 6.3
With System 7's
Macintosh File Sharing,
each Macintosh can be an
AFP client or seivei.
folders on your local hard disk that you wish to make available to
certain users or groups. Figure 6.3 shows several of the Macintosh
dialog boxes, under System 7, that control or monitor File Sharing.
With Macintosh File Sharing, there is no dedicated, centralized file
service. Each participating user acts as a client and as a server.
Other AFP Clients and Servers
The AFP standard is not limited to the Macintosh platform. In fact,
Apple made sure that AFP was an open standard that could be
implemented on other computers. As shown in figure 6.4, there are
AFP servers available for most popular platforms, including DEC
VAX, IBM PC, UNIX, and even dedicated hardware boxes.
For the most part, AFP services on platforms other than Macintosh
appear to the user exactly like a Mac running the AppleShare or
File Share software (although there may be some differences with
filename limitations, or the number of nested folders). An example
of this is found in the VAXshare component of PATHWORKS for
Chapter Six Macintosh Senaces
Macintosh, from Digital. VAXshare is software that enables a
VAX/ VMS computer to appear as a AFP server (or servers). The
AFP volumes, or disks, can correspond to any VMS directory.
Macintosh folders on these volumes correspond to VMS sub-
directories.
Macintosh Client UNIX AFP Server
Figure 6.4
AFP Servers aiefl'llimileil
loMatirtlosh.
One simple problem that often happens with mixed-platform
environments is filenames. VMS filenames cannot contain spaces,
so VAXshare must maintain a conversion map that translates
Macintosh filenames into acceptable VMS filenames. Another
problem might occur if the VAX is also used to store and share DOS
files. Wliile the Macintosh and VAX/VMS environments support
long filenames, DOS filenames are restricted to an eight-character
limit. Therefore, if you intend to share Macintosh files with DOS
users over a shared filing system, you may need to restrict your
Mac filenames to eight characters.
Print Services (PAP)
For many Macintosh users, the act of network printing is not
normally viewed as a client/server activity. In fact, the print
services provided over AppleTalk netw^orks are an excellent ex-
ample of client/server computing. Wliile the Macintosh supports
Part Two Macintosh Networking
many printers and printing protocols, the most common print
service is provided by Apple’s Printer Access Protocol (PAP).
Figure 6.5
KetPICIol an Apple
laserWrilei.
Apple LaserWriter Family
Perhaps the most popular Macintosh client/server application is
the printing of documents to an Apple LaserWriter printer. Just as
in the case of the dedicated AppleShare file server, the Apple
LaserWriter is an example of a client/sei"ver transaction, where the
client Macintosh identifies and selects a networked print server,
then sends the print job to the printer for subsequent queuing and
printing. Figure 6.5 shows the corresponding NetPICT of a
LaserWriter.
Macintosh Client LaserWriter Server
As was the case with AFP, a special protocol called the Printer
Access Protocol (PAP) was developed to handle the unique and
specific requirements of networked laser printing. PAP manages
the printing specifics, provides the queuing, downloads fonts when
required, and even informs die user if the printer is out of paper or
if the paper tray is out of the printer.
Other PAP-Compliant Printers/Spoolers
PAP print services aren’t limited to Macintosh. They also can be
found on UNIX and VAX/VMS hosts as well. There are numerous
products that allow these computers to accept and spool Macin-
Chapter Six Macintosh Services
tosh print jobs. Spooling relieves the Mac of the burden of locally
spooling print jobs on its hard disk. This practice causes the
annoying delays seen when the Print Monitor application is used.
Instead of local spooling and continual polling to check the
printer’s status, the remote PAP spooler accepts the print job as
fast as the Mac and network can send it. Figure 6.6 shows a
NetPICT of the PATHWORKS VAXshare PAP spooler running on a
DEC VAX. To the Macintosh client, the spooler program appears
exactly like a LaserWriter. The program queues the print job and
forw'ards it to the LaserWriter.
This program, or This program, or
vlrtuai LaserWriter, virtuai Macintosh,
appears to the Mac appears to the reai
user as a reai LaserWriter as a
LaserWriter. reai, printing, Mac.
Figure 6.6
Flint spooling under
DEC'S PAIHWORKS lor
Mociniosh.
Apple Open Collaboration
Environment (AOCE)
AOCE is a new set of Apple services that integrates a wdde variety of
services necessary for collaborative activities. These services will
let the Macintosh manage personal communications, workgroup
collaboration, and enterprise-wide workflow. An important part of
AOCE is the extensible directory services that will provide personal
or distributed listings of users, network services, and even phone
numbers.
Part Two Macintosh Networking
Network security is always a concern, and AOCE provides tools
that ensure private and secure communications. A new AppleTalk
protocol, the AppleTalk Secure Data Stream Protocol (ASDSP),
encrypts network traffic to prevent packet snooping. Another
technique offered by AOCE, digital signatures, will provide a way
for users to sign electronic documents. These electronic signatures
will be tamperproof and provide users the assurance that their
messages and authorizations are secure.
Starting in 1993, AOCE will lay the foundation for a new generation
of shared, cooperative applications. Messages such as voice, fax,
mail, and even video will be managed through a consistent
interface.
Terminal Services
Although terminal emulation is a throwback to an earlier time,
there are still a number of application services that require the use
of terminals. The Macintosh offers a number of terminal emulators
that work with DEC, UNIX, IBM, Prime, Data General, and many
other hosts. The types of terminals supported include most of
Digital’s VT series (that is. the VTIOO, VT220, VT340...), IBM 3270
and 5250 terminals, Tektronbc graphics terminals, and many
others. These emulators are able to connect with direct serial
connections, or through network connections such as LAT, TCP/
IP, DECnet or SNA. Figures 6.7 and 6.8 use the NetPICT diagrams
to illustrate the difference between a terminal emulator (VT220)
running over a conventional serial connection and a network
connection using LAT over Ethernet. A more detailed description
of terminal services can be found in the platform-specific chapters
later in the book.
Chapter Six Macintosh Services
VT Terminal Host Computer
VT Terminal Emulator VAX Host Computer
Figure 6.7
AVIZ20conneciedovera
seiialconneciionllie
serial coiamanicaiions
liae is a single-use
tonneciion only. Unlike a
neiwnrkcnnneclinn.il
cannot be sM.
Figure 6.8
AVI220connecieiloveia
lAI network. leiminal
access through a network
connection can be shareil
by many users.
In the past, most termined services were provided over serial
transmission lines (RS-232) that used an asynchronous protocol.
Today, terminal services are often delivered over LAN connections
and protocols. Figure 6.8 shows terminal access with DEC's LAT
protocol, which is part of their PATHWORKS for Macintosh
product.
Alternatives to conventional terminal emulation include those
products that create a front end to terminal-based services. As an
example, products from Apple, Avatar, and DCA offer program-
ming interfaces so that programmers can build Macintosh applica-
tions that have the look and feel of a Mac application, but utilize
terminal traffic as a communications medium.
Part Two Macintosh Netu'orking
Other examples that build atop a terminal session are
MacWorkStation from United Data Corporation and MitemView
from Mitem. With MacWorkStation, a host application sends
simple ASCII codes to a Macintosh. The Mac interprets these codes
as commands that control various aspects of the Mac interface.
MitemView uses standard terminal messages or MacWorkStation
codes, together with HyperCard, to act as a front end to the host.
Data Access Language (DAL) and
Dther Database Services
DAL’s purpose is to enable Macs (and other DAL clients, such as
PCs) to uniformly access relational databases (see figure 6.9).
Extending the SQL language, DAL provides a standard that can be
used on different computing platforms and different databases.
Figure 6.9
NeiPICIolaClieni/Seiver
DAUelalionship.
Macintosh Client
Host Computer
The DAL software on the client side is implemented as an optional
extension for System 7 (it used to be part of System 7 software, but
Apple recently unbundled it). DAL client applications can be
written as Mac applications or as HyperCard stacks. These stacks
contain the DAL XCMDs that give the HyperTalk programmer
access to the necessary DAL commands. In addition to HyperCard,
Chapter Six Macintosh Services
DAL is supported by other development environments, such as
Andyne’s Graphical Query Language (GQL) and the iconic, object-
oriented development environment from Serius.
There are also many Mac applications that have built-in DAL client
capabilities. Spreadsheets such as Microsoft Excel and Lotus 1-2-3
are DAL-equipped and can be used to access and retrieve data
managed by relational databases. Clear Access, from Fairfield
Software, is a desk accessory that can be used to construct ad-hoc
database queries.
DAL servers are currendy sold by Apple, DEC, Tandem, Novell, and
Pacer Software. Apple offers DAL servers for DEC VAX/VMS, IBM
mainframes, IBM AS/400, and Macs running A/UX (Apple’s version
of UNIX). The VAX/VMS version supports DEC’S Rdb, Ingres,
Oracle, Sybase, and Informix databases. The IBM mainframe
version supports IBM’s DB2 and Teradata’s DBC/1012 databases.
Pacer offers a DAL seiwer for HP UNIX computers, DEC Ultrix, and
Sun’s SPARCstation. Tandem has a DAL server for their line of
computers, and Novell has added DAL support for NetWare’s SQL
NLM.
The appeal of DAL is that the client applications are insulated from
the host database. This makes it easy to switch the database and
host server as needed. DAL also provides a single standard so that
software developers can offer database access without having to
deal with each database vendor’s own proprietary access language.
Since DAL uses a common-denominator approach to database
connectivity, some specialized features, such as database triggers,
may not be available through DAL.
Other products get around this limitation. SequeLink (by
TechCnosis) doesn’t rely on DAL; instead, it uses specific database
connections for each host database. Database companies, such as
Ingres, Oracle, and Sybase, have also gotten into the act by devel-
oping Macintosh client tools that work with their specific database
Part Two Macintosh Networking
Figure 6.10
NeiPICI of an Oracle
database conneciion.
languages. Figure 6.10 shows an Oracle database being accessed by
a Macintosh client running the Oracle client software.
Macintosh Client Host Computer
Mail Services
Perhaps more than any other Macintosh service, E-mail has the
potential for truly transforming an organization. Users can finally
quit die annoying game of telephone tag and enhance their
communications with their co-workers. E-Mail can be used as a
communications framework where voice, textual messages, and
binary file attachments are sent and tracked over the entire net-
work. In fact, many companies are now starting to view E-mail as
the engine that will power the drive toward the elusive goal of
workflow automation. Using custom-designed electronic mail
forms, a company can replace its current paper-based operations
with a system that eliminates many of the inherent shortcomings
(time delays, manual records keeping, loss of data, transcription
errors, and so forth) of the past.
During the past five years, support for Macintosh E-mail has
become widespread. As mentioned before, Apple’s AOCE initiative
will set the foundation for a new generation of mail applications,
where various message formats (mail, fax, voice, and video, for
Chapter Six Macintosh Services
example) are integrated and presented to the user within a consis-
tent interface. AOCE will also provide the security and authentica-
tion required by collaborative applications.
Today, the Macintosh is supported by most of the popular
multiplatform E-mail systems. Two of the most popular Macintosh
mail programs are Microsoft Mail for Macintosh and CE Software’s
QuickMail. Both of these products offer client and server compo-
nents that work in an AppleTalk network environment (see figure
6.1 1). The challenge begins when other mail systems, multiple
platforms, and different transport protocols are used.
Mac Mail Client
Mac Mail Server
MS Mail
CE QuickMail
MS Mail
CE QuickMail
Formats \ /
Formats
AppleTalk 1 [
AppleTalk
. Cabling ^ y
Cabling <
Figure 6.1 1
Applelalfbasedmail
services lot ilie
UaciaiosI).
Many Macintosh mail vendors offer maU gateways for their prod-
ucts that are used to meld different mail formats or transport
protocols. These maU gateways are similar to the other transport
gateways mentioned earlier in this book. A mail gateway deals with
all aspects of the mail message. This includes the formatting of the
message as well as any changes required in the transport protocol
used to deliver the message.
The most common Mac and PC gateways are used to connect to
these LAN mail systems to the UNIX mail standard SMTP (which
stands for Simple Mail Transfer Protocol) or to the emerging
industry-standard of X.400. Other less-common gateways provide
Part Two Macintosh Networking
Figure 6.12
A Maciniosh mail server
tofaiigaiewaif.
access to other mail environments, such as IBM PROFS, DEC All-
In-1, MCI Mail, and Novell MHS. There are even mail gateways
that forward mail messages to a networked fax server (see figure
6.12)1 These mail gateways run on the computer that’s running the
mail server application, or on a separate computer that’s con-
nected to the mail server.
Figure 6.13
A Macinlosli mail server
toVAXgaieivav.
As an example of a shared mail server and gateway, figure 6.13
shows a Macintosh QuickMail server that is also acting as a gate-
way to VMS Mail running on a DEC VAX computer. In this case, the
Macintosh users are using the AppleTalk protocol to communicate
with the server, and the server is using the DECnet protocol to
communicate with the VAX. The product that delivers this gateway
capability is Alisa’s MailMate QM.
Macintosh E-mail systems that support multiplatform access can
be very complex — they probably deserve a separate book to
adequately describe their capabilities. Hopefully, the descriptions
Chapter Six Macintosh Services
and diagrams found in Live Wired wiW place the elements of these
systems into a clear and comprehensible form, opening the door to
future discovery.
Conclusion
Of course the services listed previously only scratch the surface.
The Macintosh offers hundreds of networkable applications, such
as time management, resource management, backup/archival, and
even games that work over AppleTalk networks. Since the Mac was
designed at the outset to operate in a networked environment,
most Mac applications naturally take advantage of network capa-
bilities. Today, v\ath the System 7 features of Publish and Subscribe
and AppleEvents, the Macintosh has gone beyond mere client/
server applications, using tlie AppleTalk network to automatically
update shared objects and to enable applications to directly
control and manipulate other services available on the network.
Macintosh
Formats
ften overlooked or underestimated, the data
format is the single most crucial aspect of the
network. More important than cabling or proto-
cols, the formats and data structures that encode
our file access methods, word processing docu-
ments, spreadsheets, drawdngs, sounds, and multi-
media presentations represent the true measure of success when it
comes to network interoperability.
The binarjf underpinnings of all computer formats are platform-
independent. All computers express information in the same
way — with binary numbers. The only difference is that computers
and programs have different coding schemes to represent the data.
It’s important to realize that any computer can read, write, and
store any sequence of binary numbers. This means that any binary
computer can read, write, and store any other binary computer’s
data. The difficulties start to arise when one computer is able,
through its programming, to act on and manipulate that binary
data in a manner that another computer is unable to do. This is
one of the central problems in computing today.
Part Two Macintosh Networking
For every unique computer application or service, there is usually a
unique binary representation for its data. This is why MacWrite
documents are different from Microsoft Word documents, even
though they might have identical content. The choices for
interoperability are limited: either agree on a common data
format, or provide interchange utilities to convert one format to
the other. In all likelihood, we will always have a mbc of these
approaches. There will be continued pressure for data standards,
but as innovation continues to fuel the industry, new formats and
data structures are inevitable. Let’s look at some of the predomi-
nant data standards and interchange utilities that are available on
the Macintosh.
ASCII Text and Word Processors
As stated earlier in Chapter 2, "How Does Networking Take Place,”
ASCII text is a standard that uses eight bits of data to encode the
English alphabet, numbers, symbols, and special computer control
characters (such as line feeds and carriage returns). Although text
editors and word processors offer the ability to change the dis-
played font of an ASCII text file, the ASCII format doesn’t support
fonts or special attributes such as bolding or underlining. There-
fore, using ASCII as an exchange medium will only preserve the
text; any font information or special formatting will be lost. All
Macintosh word processors and spreadsheets can read and write
ASCII text. Many other Macintosh applications can also read and
write ASCII text. For example, HyperCard can read and write ASCII
text through its HyperTalk scripting capability. Most Macintosh
programming environments, such as Apple’s MPW and Symantec
Lightspeed C, also use ASCII text files for source code files.
Chapter Seven Macintosh Formats
Macintosh word processors usually rely on proprietary formats for
their documents. Word processors that are available on multiple
platforms usually use the same file format for all platforms, or
contain built-in translators that convert one platform’s format to
the other. Microsoft Word has translators that translate documents
between the Mac and Windows versions of Word. Word also has
built-in translators for other competing Macintosh word proces-
sors, such as MacWrite and WordPerfect. Figure 7.1 shows the wide
range of formats supported by Microsoft Word 5.1.
rSaue File as Type-
Normal [▼
v^Normal
leHt Only
Te»t Only with Line Breaks
Microsoft Mac UHord 3 .h
I nterchange Format (RTF)
Stationery
MaclOrite
MacLUrite II I.h
Reuisable Form Tent (RFT-DCfl)
Tent with Layout
Ulord for MS-DOS
ILIord for UMndows 1
Ulord for lilindows 2.0
WordPerfect 5.0
WordPerfect 5.1
Works 2.0 for Macintosh
Figure 7.1
MiciosoliWoiil can read
andwriieanumbefol
dillerenifofinais.
Claris uses a technology' called XTND that can be used to translate
between different word processing formats. With XTND, for
example, MacWrite users can read and write Microsoft Word
documents using options found in the Open and Save As dialog
boxes.
If a word processing document is truly foreign (Wang, for ex-
ample), then a dedicated conversion program such as MacLinkPlus
from DataViz is often required. MacLinkPlus supports hundreds of
formats from many different platforms.
Part Two Macintosh Networking
MacPaint and PICT Formats
The Macintosh introduced a number of new graphic formats.
MacPaint was one of the first Macintosh applications. It intro-
duced paint programs and bitmaps to the world. The MacPaint-
style bitmap data structure is still in widespread use today. The
original bitmap definition was 72 dots per inch and one-bit color
(either black or white). Today, Macintosh bitmaps exist in a range
of resolutions and millions of colors.
Another Macintosh format is known as PICT. The PICT format is an
outgrowth of the Macintosh imaging environment. The Macintosh
System contains a programmers’ toolkit knowm as QuickDraw. It is
a two dimensional, integer-based world where programmers can
create lines, arcs, circles, rectangles, text, and Mac bitmaps. The
QuickDraw data structures are used to display these objects on
the Mac monitor, and to print them to QuickDraw-based printers
such as the Apple StyleWriter II.
A PICT file is a series of QuickDraw commands organized in a
special resource. Thus, PICT files are limited to the objects and
precision found in QuickDraw. Because PICT files are so intimately
linked to the display environment of the Mac, the PICT format has
become a standard for the storage and management of graphics.
There are two implementations of PICT: the original, and PICT2
(which contains color and grayscale support).
Because the QuickDraw standard supports both object-oriented
graphics (that is, lines, arcs, rectangles, and text) and bitmaps, die
PICT standards support these elements as well. A PICT file can
contain only objects, only bitmaps, or a mixture of both. One
potential problem with PICT is its limited precision. As mentioned,
QuickDraw is integer-based. Under certain circumstances, when
PICT files are scaled or resized this limited precision can cause a
certain amount of distortion. This problem doesn’t occur with
Chapter Seven Macintosh Formats
PostScript images, because PostScript uses a higher precision
floating-point world.
Since Macintosh has made inroads into the graphic and publishing
worlds, the PICT standard has become an industry standard; many
drawing, illustration, and publishing applications on other plat-
forms include PICT support.
PostScript
When Apple introduced the LaserWriter, it also introduced Adobe's
PostScript imaging language. PostScript is part programming
language and part data sructure. PostScript instructions can be
used to drive laser printers, display graphics on monitors, or to
serve as a graphical format suitable for inclusion in published
documents. PostScript soon became a standard in the computing
world, both as a printer language and as a display language. While
other companies (most notably NeXT) have adopted PostScript as
a screen display language, Apple has continued to support and
enhance QuickDraw. Because of this, Apple has two imaging
standards; QuickDraw for the display and certain printers, and
PostScript for other printers (see figure 7.2). PostScript, like PICT,
can contain lines, arcs, rectangles, text and bitmaps. It uses float-
ing point numbers to locate objects; this provides a higher degree
of precision.
A variation of PostScipt, called Encapsulated PostScript or EPSF,
has become another industry standard for graphic files. EPSF files
are essentially PostScript files with the addition of “bounding box”
data that defines the size of the graphic. Most Mac illustration,
drawing, and publishing programs are able to read and write the
EPSF format.
Part Two Macintosh Networking
Figure 7.2
A Maciniosli can generate
eiilierQeickDrawnr
PnsiSciigiouigui.
Apple StyleWriter II Macintosh LaserWriter Server
IMPORTANT NOTE: One word of caution about EPSF files. There are
two versions: a generic version, and a Mac-specific version that contains a
speciai PICT preview file resource. This additionai resource, sometimes
referred to as a "thumbnail," is used to provide an onscreen display image
which approximates the real PostScript image. When the file is finaliy printed,
the reai PostScript is used.
With many Mac appiications, if you insert an EPSF file without the PICT
preview, you’ll only be able to see a rectangle that represents the border of the
image. The image will still print properly. Some applications can convert
generic EPS files into Mac-specific EPS files, and DataViz also provides a
MacLinkPlus translator for this purpose.
TIFF and GIF
TIFF and GIF are raster formats (that is, bitmap formats) that are
commonly supported on many platforms. T/FF (Tagged Image File
Format) is a standard format that’s frequently used by digital
scanners. TIFF files are often used to store high-resolution images
that have been scanned from photographic sources. GIF (Graphic
Chapter Seuen Macintosh Formats
Interchange Format) is similar to the TIFF format, but it’s popular-
ity seems to be limited to storing images of bikini-clad females on
bulletin board services.
Because both TIFF and GIF are raster formats, they are usually
viewable and editable with Macintosh paint/bitmap programs
such as Adobe Photoshop. There are also numerous viewing and
conversion utilities (Macintosh and PC) that are able to handle
these formats.
Binary
By far the largest category of Macintosh formats are those that rely
on a proprietaiy' binary' structure. Most Mac applications use their
own binary format to store data. As more vendors start to offer
cross-platform versions of their programs, the trend of binary
compatibility is likely to increase. PageMaker for the Mac uses the
same file format as PageMaker for Windows, so it’s simply a matter
of transferring the document from one machine to the other. This
can be done with a floppy disk, or over a network. Figure 7.3 shows
a NetPICT of a PageMaker file that is shared between a Mac and
PC. The real problems start when you try to exchange documents
between incompatible applications.
Macintosh Windows/Intel PC
Figure 7.3
PaoeMakerusesihe
same binary (lie fotiaai
on both Mac and
Windows versions.
Part Two Macintosh Networking
Document Interchange
The problem can be simply stated: Most application developers
design a unique binary file format to represent and store the data
created within their applications. These formats are designed using
many criteria, which include file size, application performance,
and memory requirements. Formats are rarely designed with the
goal of inter-vendor binary compatability. This is due to valid
technical reasons, as well as those pragmatic business concerns
associated witli true binary cross-vendor compatibility. However,
there are many initiatives from vendors and industry standards
organizations that attempt to bridge the long-standing language
barrier betw'een different computer applications.
Figure 7.4
MacODA supports (lie ISO
OpenDocuiuent
ArcItiieclureSianilaiil
whiclisiapilaiilires
comppupd documents.
MacODA
MacODA attempts to solve the problem of document interchange.
It does this not through conversion but through a standardized file
format. MacODA is based on the Open Document Architecture,
which is an ISO standard (standard #8613, to be precise) for the
interchange of compound documents. (A compound document \s a
document that contains fonts and graphics, both bitmap and
object.) ODA preserves the formatting and structure of documents,
so chapters, paragraphs, headers, and footers are maintained.
Figure 7.4 uses NetPlCTs to illustrate ODA use as a common
interchange format.
Macintosh Foreign Host
Chapter Seven Macintosh Formats
The ODA standard specifies three implementation levels.
# Level 1: Textual data
# Level 2: Textual and Graphical (word processing)
# Level 3: Textual and Graphical (desktop publishing)
At the present time, Apple’s (and most other vendors’) implemen-
tation of ODA supports Level 2, so effective transfers are limited to
basic word processing documents. For example, at this time,
MacODA only supports the Helvetica, Courier and Times type-
faces. The OSI standards continue to evolve and it’s expected that
MacODA will be continually enhanced.
Adobe Acrobat and PDF
Another document interchange standard is being developed by
Adobe Systems. With Acrobat, they plan to extend their PostScript
standard to support a new file format called the Portable Docu-
ment Format (PDF). PDF will support fonts, graphics, text and
color in a platform-independent file format. PDF, like ODA, will be
a common interchange format, supported by Macs, PCs, and
LFNIX. (A NetPICT of the PDF standard is shoum in figure 7.5.)
Initially, PDF will be a view-only format; users won’t be able to edit
the information contained in a PDA document. Adobe may add
this capability sometime in the future.
Macintosh Foreign Host
Hgure 7.5
IhePDf Standard fiom
Adobe promises a unilied.
platform independent
interchange lormat
Part Two Macintosh Networking
Conversions
We have just discussed one approach to document interchange;
using a mutually agreed-upon standard that is supported by many
applications. The other approach is to simply convert one format
into another.
Both approaches have their tradeoffs. With the common format
standard mentioned before, each application must support every
element and entity in the common format. This places quite a
burden on the common format, as it must accept or adapt to a rich
superset of elements used by all participating applications. This is
why common file standards are so difficult to standardize. The
benefit of this common format is that each vendor need only
implement a single converter, or translator, for their application.
The other approach is through conversion. Here, each format is
explicitly converted into another. The advantage is that the conver-
sions can be explicitly tailored and tweaked to accommodate the
two participants. Explicit conversion often provides a greater
degree of conversion accuracy. The problem is that tlie number of
required conversions begins to multiply with each aditional
format. We'll now look at two examples of utilities that facilitate
the direct, seamless conversion between different applications.
Claris XTND
Claris's solution to the problem of file translation is an open
exchange architecture known as XTND (pronounced “extend”).
XTND translation has been offered with MacWrite II, and with
most other Claris applications as well. It has also been adopted by
other software developers because Claris has licensed the technol-
ogy. With XTND, additional translation modules are simply added
to a folder (figure 7.6), after which they appear in the Open and
Save As dialog boxes (figure 7.7). XTND is also used by those Claris
Chapter Seven Macintosh Formats
applications that offer a Place command. For example, you can
place an EPSF or TIFF file into a MacDrawPro document; the
appropriate XTND translators automatically convert and import
the graphic.
^ Claris Translators =
9 items
107.4 MB in disk
7.9 MB availa
Claris XTND Bridge
PICT
MacPaint 2.0
1^
MacOraw IM.l
EPSF
Plain Text
MacWrite II
TIFF
MasVrite 5.0
9
Figure 7.6
XIND filters.
Figure 7.7
XIND Open/Save options.
Claris' XTND architecture is modular. Each translator is a separate
document that can be added or deleted to the “Claris” folder as
required. Once the XTND translators are added, they automatically
appear in the Open and Save As dialogs of XTND-compatible
applications.
Part Two Macintosh Networking
Apple Easy Open
Easy Open is a new Macintosh developer’s toolkit that incorpo-
rates seamless file translation into the operating system of the
Macintosh. It provides a standard interface for translation software
and alternate application support at the System level. If you’ve
been a Mac user for some time, you’ve probably seen the plain
document icon that produces an “application not found” message
when you attempt to open it. Easy Open will solve this problem by
coordinating the conversion of this document into another form,
or by selecting an alternative application that is capable of opening
the file.
The translation features of Easy Open are not limited to file access.
The Claris XTND environment will be managed under the auspices
of Easy Open. It will also provide transparent data conversion
during copying, pasting or publishing (System 7 Publish and
Subscribe) operations. Easy Open also adds useful file descriptions
and informative color icons to replace the annoying generic
document icons. Expect to see Easy Open services used by general
applications and specialized conversion utilities by mid- 1993.
Conclusion
Perhaps more than any other computer company, Apple Com-
puter has addressed the issue of data format portability. Starting
with the Macintosh Clipboaid, Apple has continued to provide
mechanisms to seamlessly share information between applications
and platforms. Unlike its rivals, the PC-compatibles, which have a
myriad of application-specific file formats, the Macintosh has a
few key formats tliat are supported by many applications.
Macintosh
Transport:
AppleTalk
he Transport layer is where the networking
protocols reside. Apple’s networking protocol is
called AppleTalk. It consists of many separate
protocols, each performing a specific function or
task. This chapter explains the various compo-
nents of the AppleTalk protocol family and
discusses why it is destined to be one of the most
important networking protocols for the future.
The AppleTalk Protocol Family:
An Overview
AppleTalk, the networking protocol, actually consists of many
separate protocols that work in conjunction to deliver services to
the Macintosh user. This collection of protocols is often referred to
as the AppleTalk suite of protocols.
Part Two Macintosh Networking
Figure 8.1
Ihe AppleTalk pioiQcol
slack.
AppleTalk and the 7-Layer OSI Model
The easiest way to describe and explain the AppleTalk protocol
suite is to use the 7-layer OSI model. Each AppleTalk protocol
exists at a certain layer in the stack. Sometimes, there is more than
one protocol at a given OSI layer. The AppleTalk protocols span the
layers from the Presentation layer down to the Data Link layer of
the model. Remember from the printing example at the end of
Chapter 3, “Network Diagramming with NetPICTs” that each layer
in the stack has a client/ser\'er relationship with its neighbor.
Figure 8.1 shows the entire suite of AppleTalk protocols, using the
7-layer OSI model.
Let’s take each AppleTalk protocol and describe its basic function,
starting at the top and working our way down to the bottom.
Unless you plan to develop and program AppleTalk applications,
it’s not important to understand in detail each of these protocols.
Instead, it’s only necessary to understand the basic function of the
protocol and where it fits into the larger scheme.
0 ,
Chapter Eight Macintosh Transport: AppleTalk
Presentation: APR PostScript and QuickDraw
As mentioned earlier, the OSI Presentation layer is the same as the
NetPICT Format layer. At this layer, as shown by figure 8.2, the
protocols deal with the format of the data. For example, Apple
developed the AFP protocol to address the needs of shared file
service. PostScript and QuickDraw solve the problem of device-
independent imaging.
Application
Presentation
Session
Transport
Network
Data Link
Physicai
Figure 8.2
Ilie Apple Filing Pfoiocol
lAFPiPoslScripl. and
OuickDiawaieall
eiamplesolApplelalk
Preseniatipnlayei
piDlocols.
Session: ZIP, ASR PAR ADSR and ASDSP
At the Session layer, the Macintosh uses ZIP, ASP, PAP, ADSP and
ASDSP (see figure 8.3).
Part Two Macintosh Networking
Figure 8.3
MacinlosI) Session layer
proiocols.
Application
Presentation
Session
Transport
Network
Data Link
Physical
ZIP stands for Zone Information Protocol. Its purpose is to deter-
mine which networks are associated with which zones. To accom-
plish this task, ZIP is used by the AppleTalk routers on the network
to maintain a Zone Information Table (or ZIT). This table, exempli-
fied by Table 8.1, maps networks with their corresponding zone
names. The table is stored within each AppleTalk router on the
network. ZIP also has the facility to make changes to the table
when the configuration of the network is altered. A good example
of the use of ZIP is the Macintosh Chooser. When the Chooser is
opened on a network with zones, ZIP is used to get the current list
of zone names from the router.
Chapter Eight Macintosh Transport; AppleTalk
Table 8.1 Zone Information Table
Network Number
Zone Name
10
Philadelphia
20
Wilmington
21
Philadelphia
22
Dover
119
Paris
ASP, or the AppleTalk Session Protocol, maintains a logical network
connection between an AppleTalk client and server. ASP is respon-
sible for starting and stopping each session, as well as maintaining
sequencing. ASP can be likened to the page numbers of a docu-
ment. If I were going to send you a document by copying every
page and sending them one at a time, you would use the page
numbers to reassemble the document in tire correct sequence. The
page numbers would also eliminate the potential of missing or
duplicated pages. For example, if I sent you a 20-page document,
and you received two copies of page 5 and didn’t receive page 6, it
would be an easy matter for you to discard the extra page and
request page 6 from me.
Session management and sequencing control is crucial to main-
tain a reliable connection between network partners. ASP doesn’t
provide a service directly to the user; instead, it’s used by the
higher level protocols, such as AFP, to keep things in order.
PAP is an acronym for Printer Access Protocol. It manages commu-
nications with network printers, usually LaserWriters. PAP is
sandwiched between the Presentation layer of PostScript and die
Transport layers of ATP and NBP. Basically, the function of PAP is
to open, maintain, and close a network session with the printer.
Part Two Macintosh Networking
and to send the print instructions (the PostScript code). PAP is also
responsible for determining the printer’s status. These printer
status indicators are often displayed in the Print Monitor, or in a
print status window that appears across the top of the Macintosh
screen after print invocation.
ADSP is the last of the Session protocols. It stands for Apple Data
Stream Protocol. Because ADSP is a Session protocol, part of its job
is to open, maintain, manage and close connections between
tw'o network devices. In addition, ADSP provides for efficient,
bidirectional delivery' of data vv'ithout loss or duplication. Unlike
the other Session layer protocols, which connect to various Trans-
port layer protocols one layer down, ADSP skips a layer and
connects directly to the Netw'ork layer of DDP. This is because
ADSP duplicates some of the functionality of the Transport layer. A
new alternate version of ADSP, called ASDSP (Apple Secure Data
Stream Protocol), has recently been added to the AppleTalk suite.
It is part of the Apple Open Collaboration Environment (AOCE)
and provides secure communications through encryption tech-
niques.
In some ways, the combination of PAP and ATP (discussed next)
provide functionality similar to tliat of ADSP (and ASDSP). PAP and
ATP work in concert to provide a reliable, sequenced, and man-
aged connection to printers: ADSP (and ASDSP) provide the same
capability for general netw'ork connections.
Transport RTMR AURR ATR NBR and AEP
At the Transport layer, the Macintosh uses RTMP, AURP, ATP,
NBP, and AEP (see figure 8.4).
RTMP stands for Routing Table Maintenance Protocol. RTMP is a
protocol spoken by AppleTalk routers, designed to keep AppleTalk
network routing tables current. Briefly, an AppleTalk routing table
is a listing compiled by a router, consisting of each AppleTalk
Chapter Eight Macintosh Transport: AppleTalk
network, its corresponding distance, which router port is used to
access that network, and the AppleTalk node number of the
nearest router used to access that network. RTMP packets are
exchanged by the routers on a regular basis.
Application
Presentation
Session
Transport
Network
Data Link
Physical
Figure 8.4
MacintostiTranspofi
layer proiocois.
RTMP is about to be supplanted by AURP. AURP, the Apple Update
Routing Protocol, is a new routing protocol that only sends out
routing tables when a network change occurs. This new protocol,
First introduced wdth the Apple Internet Router 2.0, will make
AppleTalk better suited for use over wide-area networks. Since
routing is so important in an AppleTalk network, RTMP and AURP
will be discussed in greater detail later on in the chapter.
ATP, the Apple Transaction Protocol, provides reliable transport
services between the source and destination sockets. Three differ-
ent types of ATP packets ensure this delivery. First, a transaction
Part Two Macintosh Networking
request gets the attention of the destination socket. The destination
socket replies back to the source witli a transaction response, and
the source finishes the transaction with a transaction release. As an
example, ATP is used by PAP to provide reliable printing to a
LaserWriter. It’s as if the Mac was saying, “Here’s some PostScript,”
followed by the LaserWriter saying, "OK, 1 got it,” followed by the
Mac saying, “OK, here’s some more.”
NBP is the Name Binding Protocol. It’s purpose in life is to link
names (such as "Joe’s LaserWriter” or “Bob’s Mac”) to AppleTalk
addresses. This protocol will be discussed in depth when we
analyze how the Chooser works.
AEP stands for the AppleTalk Echo Protocol. This is a simple
protocol that is used for diagnostic and testing purposes. AEP can
be used for two purposes: to check for the presence of another
node, or to get an estimate of the round-trip delay time between
two nodes on the network. A prime example of AEP is Apple’s
Inter»PolI utility. It can be used to send a variable number of echo
packets to another device on die network. The minimum, maxi-
mum and average transit times are displayed.
Network: DDP
At the Network layer, the Macintosh uses the Datagram Delivery
Protocol to address the message (see figure 8.5).
The Datagram Delivery Protocol, or DDP, is the primary protocol at
AppleTalk’s Network layer. DDP places AppleTalk communica-
tions into a suitable "container” and addresses them for delivery
on the internetwork. All the higher-layer data is encapsulated
within a DDP datagram.
The datagram is labeled with the source and destination AppleTalk
address. This means that the datagram will have two 32-bit
AppleTalk addresses that contain the network, node, and socket
Chapter Eight Macintosh Transport: AppleTalk
numbers of the sender, or source, and the recipient, or destination.
When routers route AppleTalk packets, they look at the source and
destination network numbers in order to determine the best way to
route the packet.
Application
Presentation
Session
Transport
Network
Data Link
Physical
Hgure 8.5
Maciniosli Network layei
protocols.
The datagram is independent of the cabling. Only at the next layer
down — the Data Link layer — is the datagram placed within the
appropriate cable-specific network frame.
Data Link: LLARELAR and TLAP
The various data link protocols (see figure 8.6) correspond to the
different cabling systems supported on the Mac.
Part Two Macintosh Networking
Figure 8.6
At the Data link layer, Ili8
Macintosh uses a number
olpiDlocols,suchasllAP,
[lAP.andUAP.iohanille
dillerent kinds nf cabling.
Application
Presentation
Session
Transport
Network
Data Link
Physical
LLAP stands for the LocalTalk Link Access Protocol and is the built-
in driver that comes with every Mac and networked LaserWriter,
and with many LocalTalk peripherals. The LLAP driver takes the
DDP datagram and places it in an LLAP frame for subsequent
delivery over a LocalTalk network.
ELAP stands for the EtherTalk Link Access Protocol and is present
on Ethernet-equipped AppleTalk devices. The ELAP driver takes
the DDP datagram and places it in the data field of an Ethernet
frame.
TLAP stands for the TokenTalk Link Access Protocol, and places the
DDP datagram within the data field of a Token Ring frame.
Chapter Eight Macintosh Transport: AppleTalk
How Does AppleTalk Work?
As mentioned in Chapter 4, "Networking Concepts," AppleTalk
communication is based on the premise that all devices and
network processes are uniquely identified. For AppleTalk, this
requires network, node, and socket numbers.
AppleTalk Addressing
AppleTalk network numbers are 16 bits in length: there are 65,536
potential network numbers available for use (see figure 8.7).
Figure 8.7
Willi IB-bil network
numb8is.A|iplelalk
siipporis more Ilian
6!i.D00nelwoik$!
In the simplest of networks, consisting of a Mac and a LaserWriter
connected with LocalTalk, the network number is somewhat
useless. There are no routers to maintain the network numbers,
and therefore in this case there are no network numbers (see figure
8.8), With a single non-routed LocalTalk (or PhoneNET) network,
the network number is always 0.
Figure 8.8
Inasimplelocallalk
neiwoik.ilie network
numliei is always 0.
Part Two Macintosh Networking
In actual network transmission, the 16 bits consisting of ail zeros
aren’t even included in the source and destination of the commu-
nications. Why carry 32 bits of “nothing” with every transaction?
When these all-zero network numbers are omitted from the source
and destination of the DDP datagram, it’s known as a Short DDP.
In cases where the network number is non-zero (and is therefore
necessary), the DDP datagram is referred to as a Long DDP. Addi-
tional addressing details will be covered later in this chapter, when
AppleTalk Phase 1 and Phase 2 are described.
The Importance of Dynamic Addressing
Imagine moving into a new house. One of the first things you need
to do is to get a new phone. Normally, when you get a new phone,
the phone company assigns you a number from their registry of
phone numbers. But imagine a phone company that doesn’t want
to be bothered with the tedious administrative details of assigning
unique phone numbers.
This phone company has a system, whereby you simply dial any
phone number at random. If someone answers the phone (see
figure 8.9), you apologize and hang up. If there’s no answer, then
that number becomes your new phone number.
Figure 8.9
Someone answers
lire phone...
Chapter Eight Macintosh Transport: AppleTalk
This imaginary scenario is the mechanism that Apple uses to
dynamically assign AppleTalk node addresses. With other proto-
cols, such as DECnet and TCP/IP, the node ID number is deter-
mined and assigned by a human. This human usually has a big list
or spreadsheet of node assignments in order to determine which
numbers are available for use.
Apple tried to avoid this problem with AppleTalk by implementing
a dynamic node-addressing scheme. When a LocalTalk-connected
Macintosh boots up on a network for the first time, it chooses a
node number at random (see figure 8.10). The Macintosh has no
way of knowing whether this random number is already in use by
another node, so it sends a special packet (known as an enquiry
control packet) to the node in question. If the enquiry reaches its
destination, the node responds to the inquiring Mac with an
acknowledge control packet. Of course, this means that the number
is in use and cannot be used as the node number for the new Mac
on the network.
Figure 8.10
AMaconalocallaik
network geneiaiesiis
own node address at
random.
Therefore, another random node ID and enquiry control packet is
generated. If no acknowledge control packet is received, then the
new Mac is free to use the number. Once a Macintosh determines a
unique node ID for itself, it stores the number in the Mac’s PRAM.
(The PRAM, or Parameter RAM, is special memory that is non-
volatile — a battery is used to maintain its contents even while the
Part Two Macintosh Networking
Figure 8.11
No answer — even
iliDugliiliere'saiihone
wiiliiliereriuesieil
numlier.
Mac is turned off.) The Macintosh will use the stored node ID as an
educated first guess the next time the machine is booted. This
minimizes the node ID contention during startup.
Generally, a Mac’s node number remains with a given Mac, but
this is not alw'ays the case. Wlien a Macintosh, or any AppleTalk
node, is turned off, it’s unable to respond to the enquiry control
packets sent out by the other machines. Using the phone example,
it’s as if the receiver of the message was asleep or out of the
house — he would be unable to answer the phone (see figure 8.1 1).
Likewise, if a new Mac is added to the network, the possibility
exists that it could randomly guess — and steal — a node ID that is
locked away in the PRAM of a powered-down Mac (see figure 8.12)
Then, when this Mac is powered on, it tries to use its stored node
ID as its first guess, and the other Mac responds with the acknowl-
edgment packet. This causes the original Mac to establish a new
node ID. Therefore, it’s quite possible for your Mac to be node 44
on Monday and node 32 on Tuesday. With AppleTalk, it’s not
important that a node number be permanently associated with a
given Macintosh.
Chapter Eight Macintosh Transport; AppleTalk
Figure 8.12
lniliiscase.llieieisa
nodeil 82 onlliea 8 two(k.
buiiliaiMacisaoi
runaingandilieieloieis
uaablatfliespondiothe
eaquiiyconliolpackei.
Ihe Mac oflihaleii will
"clear the sleeping
Mac’s npdennmbei.
If the Mac is connected to an EtherTalk network, the process for
establishing a unique number is a bit different. As we discussed in
Chapter 4, "Networking Concepts,” each Ethernet card has a
unique address stored in its hardware. AppleTalk must create a
standard AppleTalk logical address and associate it with the
Ethernet hardware address. Because of this, tlie Mac uses another
protocol — the AppleTalk Address Resolution Protocol, or AARP — to
create a logical AppleTalk address, instead of using the LEAP
Enquiry technique. Upon booting a Mac onto an EtherTalk net-
work, your Mac broadcasts a series of AARP probes. The AARP
probes contain a tentative AppleTalk logical address and the
physical hardware address of the Ethernet card. After the receiving
nodes compare the logical addresses and report back any conflicts,
the Macintosh obtains a unique logical AppleTalk address.
AARP is also used to construct an Address Mapping Table (AMT),
which links the logical AppleTalk address to the physical Ethernet
address. This dynamically generated and updated table, exemplifi-
ed in figure 8. 13, contains recent AppleTalk addresses and their
corresponding Ethernet hardware addresses.
Part Two Macintosh Networking
Figure 8.13
AADPisusedloconslrucl
an Address Mapping
fable, wliicliegyales
logical Applelalk
addresses wiihpbysical
fibernet addresses.
AppleTalk Logical Addresses
AMT for 12.42
12 . 83 : 08 - 12 - 32 - 43-91
12 . 77 : 08 - 12 - 32 - 51-71
12 . 81 : 08 - 12 - 32 - 41-92
12 . 90 : 08 - 12 - 32 - 78-92
12 . 16 : 08 - 12 - 32 - 33-71
12 . 35 : 08 - 1 2 - 32 - 41 -88
12 . 09 : 08 - 12 - 32 - 67-82
12 . 132 : 06 - 12 - 32 - 91-41
In fact, this is a very important and significant feature of the
AppleTalk protocol. As computers get smaller and more common-
place, and as wireless networking catches on, dynamic node
addressing will become crucial. It would be ridiculous to have to
stop at the front door of a company that uses a wireless LAN simply
to receive a network node ID for your Newton personal digital
assistant (see figure 8.14). These node numbers don’t possess any
intrinsic meaning — they’re simply unique numbers, and it just so
happens that computers can do an excellent job of assigning
unique numbers.
Figure 8.14
I need a node number?!?
Zones
Zone names are simply names assigned to networks. They, along
with network numbers, are stored within the AppleTalk routers. In
Chapter Eight Macintosh Transport: AppleTalk
the case of LocalTalk (Phase 1 AppleTalk) networks, there can be
only one network number and one zone name per network.
Imagine a simple internet, figure 8.15, consisting of two LocalTalk
networks connected by a router. Each network has a unique
network number and different zone names.
Figure 8.15
Iwolocallalkneiwofks
conneciedwiihaiouiei.
Eacti network has a
unique neiwoiknumbei
and dilieteni tune names.
Each network has a unique network number (10 and 20), and each
network has a zone name (Left and Right). Here, the zone name
Left is used to identify the devices in network number 10; Right is
used to identify devices in network number 20.
There’s no uniqueness requirement, however, for zone names. It’s
perfecdy fine to have duplicate zone names in order to logically
organize services. Therefore, networks 10 and 20 could have the
same zone name of “CAL.” In figure 8.16, each network still has a
unique network number, but the zone names are the same.
When networks have the same zone name, similar devices are
grouped together. So, if there were two LaserWriters in network
10 and three LaserWriters in network 20, there would be
a total of five LaserWriters in the “CAL” zone. Keep in mind.
Part Two Macintosh Networking
Figure 8.1 6
Inis example, eatli
neiwoikslill has a unique
neiwofk number, but the
2une names are the same.
however, that when there’s only one zone, the Macintosh Chooser
does not display the zone list. This is true even when there are
multiple networks v\ath tlie same zone name.
Network #10
n jn n
Zone: CAL'
1 “
1 “
Network #20
r
1^3
Zone: CAL
r
Figure 8.17
Etheinet anil Men Ring
Applelalk networks can
use a range nl network
numbers.
In the case of Ethernet and Token Ring networks (Phase 2
AppleTalk), a cable segment can be assigned a range of network
numbers and multiple zone names. In figure 8.17, the zone names
are not associated with any specific network number.
4.54 1.78 10.39 2.66 4.41 7.78
Zones: Red, White, Blue
Default Zone: Red
With Phase 2 networks (described in the next section), the zone
names are defined for the entire netv\^ork segment. When a
Chapter Eight Macintosh Transport: AppleTalk
Macintosh (or other AppleTalk node) is added to the cable, it
automatically belongs to the Default zone.
The Default zone is a designated zone where AppleTalk nodes
appear by default. It’s established by choosing a zone from the list
of defined zones maintained by the router configuration software.
If there are multiple routers on the cable, as in figure 8.18, all
routers must agree on the network range, zone list, and default
zone assignment for the cable. Mismatched net numbers and zone
lists are the most frequent causes of network problems.
Figure 8.18
AlliDuieisonagiveD
segmenimusi agree on
the network raimiier
range, tones anil the
rielaoliione.
As shown in figure 8.19, you can change your Mac's default zone
assignment by opening the Netw'ork control panel and double-
clicking on the EtherTalk or TokenTalk icon. A zone list will appear,
along with a prompt to change the Mac’s zone. From then on, your
Mac will belong to that zone — until you decide to change it again.
If you’re unsure which zone your Mac belongs to, simply open
the Chooser. The highlighted zone is the zone to which your
Macintosh belongs.
It’s important to realize that the choice of zone names can affect
network performance. For example, consider a wide-area network
that connects two AppleTalk-Ethernet LANs. As shown in figure
8.20, one network is in Washington and the other is located in
Part Two Macintosh Networking
Little Rock. One could assign zone names geographically by
creating a “Washington” and “Litde Rock" zone, or functionally by
creating an “Administrative” and “Judicial” zone. However, if you
group by function, instead of geography, there is a potential for
increased traffic.
Figure 8.19
IliBdelauluone
assionmentolaMaccan
be changed by Mle-
clicking on ibeEibeilalk
orlnkenlalkiconinibe
network Cnnirol Panel.
IP Nettuork
Select an AppleTalk connection:
m
LocalTalk
Built-In
3235BI3 Remote Only
Please select
this computer's
AppleTalk zone.
[ Cancel ]
Figure 8.20
Agglelalk zones can be
set up geographically
or lunciionally. How-
ever, there can he a
perlormancedilference.
N
Net #20-50
Zone: Washington
» 1
Net #70-150
Zone: Little Rock
Net #70-1 50
Zones: Administrative
Judicial
Let’s look at an example of functional grouping. Suppose there are
three LaserWriters located in Little Rock on the Administrative
Chapter Eight Macintosh Transport: AppleTalk
zone, and six LaserWriters located in Washington on the Adminis-
trative zone. Every time I select the Administrative zone from a
Mac in Washington, AppleTalk must not only poll for the printers
in Washington, but it must also poll over the wide-area connection
for printers in Little Rock. Eventually, I’ll see nine LaserWriters in
my Chooser. The problem with this functional approach is that
large AppleTalk networks over limited-bandwidth links can bog
down with the polling traffic. If the zones are arranged geographi-
cally, then the polling traffic remains local and doesn’t burden the
wide-area connection.
NOTE: Zone names can be up to 32 characters in length. They are case-
insensitive. This means that zone “Little Rock” and zone “LITTLE ROCK” are
the same. Zone names are much easier to read, however, if the standard
sentence case is used. Spaces are significant, so be careful not to type an
extra space between words or to add an unwanted space at the end of a name.
If you’re creating a zoned network, try to establish a meaningful
zone-naming convention prior to router installation. Remember
that network numbers are for the computer’s convenience; zone
names are for the users' convenience. Your zone-naming standard
should help users find networked printers and file servers, and
should be flexible enough to support future expansion.
Zone names of “First Floor” and “Second Floor” only make sense
if your organization has a single building. Avoid the practice of
naming zones after the room number of the wiring closet where
the router is located. Zone names should also be coordinated with
de\dce name conventions (such as AFP servers and LaserWriters).
If your zone-naming convention consists of a city, building num-
ber, and floor, it’s not necessary to repeat this information at the
device level. Instead, try to include information that pinpoints the
final location or function of the device.
Part Two Macintosh Networking
AppleTalk Phase 1 and 2
In the beginning, when Apple designed AppleTalk and LocalTalk,
there was no distinction between Phase 1 and Phase 2. Wlien Apple
developed EtherTalk, which is nothing more than AppleTalk
protocols on Ethernet cabling, they simply carried the original
AppleTalk protocol used on LocalTalk networks over to Ethernet.
Apple developed AppleTalk Phase 2 to address certain shortcom-
ings that were present in the initial EtherTalk implementation.
Phase 1 EtherTalk networks were restricted to a maximum of 254
devices and were inefficient when it came to certain aspects of
routing and broadcasting. Phase 2 fixed these problems, and also
provided the first version of TokenTalk, which is Apple’s imple-
mentation of the AppleTalk protocols on Token Ring networks.
Phase 1
Phase 1 networks, as mentioned before, can only have a single
network number per cable segment. Because AppleTalk only
supports 256 potential nodes per network number, each cable
segment was therefore limited to 256 devices (see figure 8.21).
Figure 8.21
Each AppleM network
can poieniiallir support
2S6 devices. Phase I anil
PhaseZieseniediflerent
values.
Actually, the limit is 254 devices. With Phase 1 networks, node numbers range
from 0 to 255. 0 is not used and 255 is reserved for broadcasts heard by all
devices.
Chapter Eight Macintosh Transport: AppleTalk
With LocalTalk, this is hardly a problem. LocalTalk netv\'orks have
electrical and isolation restrictions that limit the number of nodes
to a recommended limit of 32 devices. When Apple introduced the
first version of EtherTalk, it too was limited to a single network
number and 254 nodes per segment. This was bad news for large
organizations that wanted to start populating their Ethernets with
Macs. This was particularly bad news for those organizations that
had bridged Ethernets that spanned many locations. They were
limited to a grand total of 254 EtherTalk devices diroughout their
entire organization (see figure 8.22). Apple addressed this problem
in 1989 with the introduction of AppleTalk Phase 2.
1.1 1.2 1.3 1.4 1.5 1.6
Figure 8.22
liwaseasyiofeaciiilte
254 node limit ol
EilieilalkPliasel,
patiiculailif will) large,
bridged liliernei
networks.
Phase 1 and 2 Network Number Assignment
Phase 2 broke the 254 node limit by avoiding the restriction of a
single network number per cable segment. Instead, the cable
(either Ethernet or Token Ring) can be assigned a range of values.
In a Phase 2 netw'ork, the network numbers range from 0 to 65,534.
Zero (0) is undefined and not used. A special range, called the
startup range, runs from 65,280 through 65,534. This leaves the
numbers 1 through 65,279 for general assignment.
Part Two Macintosh Networking
Figure 8.23
ta I Willi no routers.
Hgure 8.24
Phase) will) routers,
network numbers can be
I ibrough 05,530
Let's review the network number assignment rules.
• Phase 1 (LocalTalk, and the now-extinct first version of
EtherTalk) with no routers (see figure 8.23).
The network number is always 0. In practice the 16-bit zero
network number is omitted from the address to conser\'e
space. This is known as a S/torrDDP address.
0.1 0.143 0.14 0.31 0.54 0.88 0.53 0.71
The network number is a single number between 1 and
65,536. Each network segment, either LocalTalk or the old
EtherTalk, must have a single number assigned. No dupli-
cates are permitted.
• Phase 2 (the current version of EtherTalk and TokenTalk)
with no routers (see figure 8.25).
Chapter Eight Macintosh Transport: AppleTalk
The network numbers fall in a range between 65,280 and
65,534. This is knowm as the Startup range. There are 254
numbers in this range that are used by AppleTalk nodes as
they come onto the network.
65280.1 ► 65280.253 65281 . 253 '
65534.253
The choice of network number, like the node number, is
made at random. One Macintosh could be in network 65,288
and be node 23; an adjacent Mac could be in network 65,500
and be node 23.
NOTE: In Phase 2 networks, there is an additional reserved node number, so
there can only be 253 nodes per network. This means that for a single, logical
segment of Ethernet, there can be 254 networks each having 253 nodes. This
multiplies for a total of 64,262 AppleTalk devices per segment. If you need
more than this, an AppleTalk router will be required.
• Phase 2 with routers (see figure 8.26).
The network numbers are a range between 1 and 65,279. 0 is
not used and 65,280 through 65,534 are reserved for the
startup range. Theoretically, you could use the entire range of
numbers for a single segment, but this would be wasteful.
Instead, it makes sense to assign a modest range of numbers
to a cable. This way, growth and expansion can be easily
accommodated.
Rgure 8.25
PhaselwinDiouleis.
When no Mis aie
preseni. Phase 2 devices
assign neiwoiknembefs
intheslarlupiange —
65,2ll0-65,534.
Part Two Macintosh Networking
Figure 8.26
Phase 2 Willi Routers.
Network numlieis can be
1 ibrough 65.229.
For example, in a large organization, it might make sense to
assign ranges to the various divisional locations. Houston can
have 1-100, Dallas 101-200 and San Antonio 201-300. Then
local network managers can add routers using network
numbers in their pre-assigned range. We’ll see later how
Apple’s new routing protocol, AURP, extends this concept
even further and provides additional flexibility in network
number assignment.
Phase 2 Transition Routing
Before 1989, EtherTalk networks were Phase 1. As mentioned
before, they were limited to a maximum of 254 AppleTalk devices.
When Phase 2 was introduced, Apple chose a different Ethernet
frame format and type code. From an Ethernet perspective, this
meant the Phase 1 and Phase 2 packets could be readily distin-
guished by network devices. It also meant that Phase 1 and Phase 2
protocols could coexist on the same cable.
The coexistence of Phase 1 and Phase 2 was originally deemed as
desirable, since large organizations making the switch would find it
difficult to upgrade all at once. The only problem with this coexist-
ence technique was that the Phase 1 and Phase 2 devices could
Chapter Eight Macintosh Transport: AppleTalk
only communicate with like devices. A Phase 1 node could not
communicate with a Phase 2 node. Apple addressed this problem
with an interim solution called transition routing.
A router (either a Mac running the AppleTalk Internet Router, or a
dedicated router, such as a FastPath) could be configured as a
transition router. The transition router’s function was to accept all
Phase 1 packets and retransmit them as Phase 2 packets, and vice-
versa. This unfortunately doubled the traffic, but at least the Phase
1 and 2 nodes could communicate during the transition process.
Today, Phase 1 EtherTalk networks are rare. Almost everj'one has
converted to Phase 2. Those sites still running Phase 1 EtherTalk
are at a significant disadvantage. In addition to increasing the
number of devices and zones per segment, Apple made other
significant improvements with Phase 2.
Phase 1 Broadcasting versus Phase 2 Multicasting
Starting with AppleTalk Phase 2, Apple altered the broadcasting
mechanism (used by NBP Lookups and other AppleTalk services)
to utilize a more specific technique known as multicasting. Let’s
first review the concept of broadcasting, using Ethernet as an
example.
Often, a networking protocol (such as AppleTalk) must make an
announcement to all the devices on the network. Imagine that you
work at a large company and you need to contact someone. If you
know that person’s phone number, you simply dial the number,
but what if you don’t know that person’s number, or they don’t
answer the phone? Then, you’ll need to have that person paged
(see figure 8.27). Loudspeakers all over the company send out a
“broadcast” message. Of course, this approach is somewhat
inefficient: you're only looking for one person, but everyone in the
company has to listen to the message. This is essentially how
Ethernet broadcasting is implemented.
Part Two Macintosh Networking
likecompanifwiilepaoino
iliai is trying 10 locale
one person.
Figure 8.27
ithernei broadcasts are
Compared to other protocols, it’s probably fair to say that
AppleTalk generates more broadcast messages. In fact, in the early
days of Phase 1 EtherTalk, AppleTalk started to get a bad reputa-
tion for being excessively “chatty” for this very reason. What
annoyed some people was that these AppleTalk-generated broad-
casts were being sent to all devices on a given Ethernet — regardless
of whether that device even spoke AppleTalk protocols. Thus, the
owners of minicomputers, PCs, and UNIX workstations started to
complain that their machines were spending an inordinate
amount of time responding to broadcasts not intended for their
machines. Just imagine how annoying it would be if the same
group of people were being continually paged and you always had
to listen to the messages since you sat right next to the loud-
speaker.
The solution is to be more specific with the broadcast messages.
Imagine if the pages were made on a departmental basis (see figure
8.28). If you need to have someone paged, the phone operator asks
you for the department of the pagee. When the page is made, only
Chapter Eight Macintosh Transport: AppleTalk
those loudspeakers in the specific department are used. The rest
of the employees are spared from having to hear the irrelevant
message. This is the concept behind multicasting. Each networking
protocol that supports multicasting has a unique multicasting
address assigned. This address is similar to the department name
used in the prior example. An AppleTalk multicast is only “heard”
by other AppleTalk nodes. A TCP/IP multicast is only heard by
other TCP/IP nodes. With Phase 2, AppleTalk traffic is also more
efficient through multicasting, since Phase 2 multicasts are con-
tained within a given zone, even though there may be only one
cable segment containing other zones.
Figure 8.28
Eiheinei Multicasts are
siinilartoileeartnient-
wide paging that is trying
toincatennepersnn.
How Does the Chooser Work?
The Chooser is one of the coolest, least understood aspects of the
Macintosh. With it, a user can find and select network (and local)
services in an easy and consistent manner. The Chooser does
many interesting things behind the scenes. It dynamically gener-
ates a list of available network zones. When you choose a particular
Part Two Macintosh Networking
service by clicking on an icon, the Chooser then generates a list of
those devices that meet the selection criteria. Finally, once you’ve
selected a particular named service, the Chooser proceeds to
discover the AppleTalk address of the chosen service. (You see, the
names in the Chooser are for your benefit, while the network
addresses are for the benefit of the Macintosh.)
Figure 8.29
Unlike Applelalk, which
dynamically discovers
neiwork addresses and
services, DECnelreqaires
aspeciliclisiingol
available nodes helore
communicatiancanbe
established.
The Unknown Address
We’ve discussed how AppleTalk dynamically assigns a node
number to a device. This was a wonderful achievement, but Apple
had an additional problem to solve. Each AppleTalk device auto-
matically generates a unique node number, but initially it’s only
known by that device. Nodes don’t automaticaUy know the node
numbers of other devices. With other protocols, such as DECnet,
this problem is solved manually. Someone (usually the DECnet
administrator) uses a utility program to create a database of node
addresses. This database, exemplified in figure 8.29, is required on
every DECnet node in order for that node to communicate with the
other nodes.
Apple, desiring a plug-and-play environment, decided that this
manual creation of node lists was not in keeping with the spirit of
Macintosh. An alternative approach was developed to solve the
problem of address determination. Let’s consider a Mac that needs
to print a document to a particular LaserWriter on the network.
Chapter Eight Macintosh Transport: AppleTalk
The Mac doesn’t have any idea of the odter nodes on the network.
Therefore, it must go through a process of discovery to identify the
available services. This is done by the AppleTalk protocol knowm as
the Name Binding Protocol, or NBP.
The process is simple. When a user opens the Chooser and clicks
on a service icon (such as the LaserWriter icon), the Macintosh first
acquires a list of zones from the nearest routers with the Zone
Information Protocol (ZIP). It then sends out a NBP Lookup
Request packet. (Actually, it doesn’t simply send it out, it broad-
casts — or multicasts — the request to all devices on the cable. This
makes sense because the Mac has no idea who to send the request
to anyway.) Essentially, the NBP Lookup Request packet contains
information on the requested named service. A name in the
AppleTalk system is called a Network Visible Entity, or NVE.
AppleTalk Names and the NVE
NVEs aren’t nodes; rather, they are the services offered by nodes.
These services are identified by their socket numbers. One
AppleTalk node could support a number of services, each with a
socket number and a corresponding NVE. The NVE consists of
three components along with delimiters and some special sym-
bols. Each name (object, type and zone) is a 32 character case-
insensitive alphanumeric name. The syntax is: Object:Type@Zone.
The Object name is the name of the entity that is usually assigned
by a person. For a Mac, the object name is defined in the
Macintosh Name field of the Sharing Setup Control Panel.
LaserWriter object names are established by the Namer utility
program. Examples of object names include: Jim ' s Mac, 3rd
Floor LaserWriter, and Joe's File Server.
The Type name is the generic name of the service. Apple maintains
a registry of these names. For Macs and LaserWriters, the type
name has traditionally been the model name of the device. Some
Part Two Macintosh Networking
examples include Macintosh Ilci, LaserWriter Pro 630, and
PowerBook 1 80. Other, more generic type names include
AFPServer and PAPSpooler. Remember, since NVEs and type
names are based on sockets, it is possible (and likely) that a single
node will have several NVEs.
The last field of the NVE is die zone name where die service can
be found. Zone names, as explained earlier, are created by the
network administrator and exist within the AppleTalk network
routers.
Here are some examples of NVEs:
• Jim's Mac: Quadra 950iaoffice Zone
• The Group's Printer:LaserWriter@New York City
In addidon to the names, there are special wildcard characters that
can be applied to the NVE:
• An equal sign (=) in the Object or Type fields means all
objects, regardless of Object (or Type) names.
• An asterisk (*) in the zone field indicates the current zone of
the requesting node.
Let’s review some examples of NVEs with wildcard characters.
= : =0* means all objects, of all types, in the current zone.
= : AFPServer0* means all objects of type AFPServer in the current
zone. = : LaserWriter0Blue means all objects of type LaserWriter
in the Blue zone.
Now, with a basic understanding of the NVE, we can return to the
Chooser. When you click on the LaserWriter icon (see figure 8.30),
the Macintosh generates an NVE that corresponds to the service
and the currently selected zone. The NVE takes die form
=:LaserWriter0* .
Chapter Eight Macintosh Transport: AppleTalk
Hgure 8.30
Wlten you click on ilie
lasefWfilericon.ilie
Macinioshgeneiaicson
Nil! (network Visible
Iniilvl. shown above.
Ihis happens aiihe
Pieseniation layer.
Deciphering this NVE, we can establish that the Macintosh is
looking for all names of type LaserWriter in the current selected
zone.
This NVE is then referenced within the NBP Lookup Request (see
figure 8.31). Remembering that NBP is a Transport layer protocol,
the next layer dowm is the Network layer of DDP. Thus, the NBP
Lookup Request is then handed down to DDP.
Figure 8.31
Ihe NVE is placed within
an NBP lookup Nequest
packet Ibis takes place
in the transport layer.
Next, the DDP datagram must be addressed. The "From" part of
the address is easy. The sending Macintosh simply uses its own
network and node number. A socket number is chosen at random
from the pool of available socket numbers. The “To” part of the
address contains the network number, the broadcast/multicast
node number of 255, and a special reserved socket number of 2,
Part Two Macintosh Netw'orking
Figure 8.32
DDPDaiagram
Adtaing.Ih8NVns
placed williin an NBP
lookup Bequest packet
Iliis lakes place altlie
Tianspoii layer.
which is known as the Names Information Socket. The complete
message, in a simplified form, is shown in figure 8.32.
JL
1^1 •••Otll
NBP LkUp =:LaserWrlter@*
DDP
From: 12.33.144
To: 12.255.2
From: 12.33.144 (the Macintosh)
To: 12.255.2 (Anyone who will listen)
NBP LkUp
=: LaserWriter®*
Let’s translate this into English:
"Hey there, all you AppleTalk nodes! This is device 12.33 commu-
nicating over socket 144. I’m broadcasting to everyone in network
number 12 and I’m particularly interested in names. I’m looking
for all the names of LaserWriters in the current zone.”
The DDP datagram, which contains the NBP request, is then
placed in a network frame (see figure 8.33). This frame is then sent
out over the cable to all the devices.
Chapter Eight Macintosh Transport: AppleTalk
± i.
1
LLAP
1
NBP LkUp
DDP
LLAP
From: 12.33
To: 12.255
Figure 8.33
IlieOOPdaiagianiis
placed inside an UAP
ftane. Since Incallalk
doesn't use physical
addresses, llAP uses the
logical Applelalk
addresses.
A device might discard tlie message if it doesn’t apply, but if the
NBP request is sent to a device where the NVE matches, such as a
LaserWriter, it accepts the datagram, opens it and processes the
NBP request (see figures 8.34 through 8.40).
NBP LkUp DDP LLAP
From: 12.33
To: 12.255
Figure 8.34
The entire conienisol
the llAP Iratne is then
broadcast nnto the
cable segment
± JL '
NBP LkUp
DDP^ LLAP
From: 12.33
To: 12.255
Figure 8.35
Since the message was
broadcast over the
network, the laserWritet
land other nodes)
process the locallalk
frame.
Part Two Macintosh Networking
Figure 8.36
Ilie LaserWriter inspects
the llAPIraitie it received
from the locallalt
hroadcasL
NBP LkUp DDP
LLAP
From: 12.33
To: 12.255
Figure 8.37
Ihe LaserWriter strips off
the LLAP frame to reveai
theDDPOataoram Now
the LaserWriter knows
who sent Ihe message.
.S. X '
I «
DDP
NBP LkUp
DDP
From: 12.33.144
To: 12.255.2
Figure 8.38
Finallif, Ihe LaserWriter
uncnverediheNBP
Lookopreqoesiand
knnws that the device
12.33.144 is interested in
names.
ItMialMrwrKv
5.. A.
1
NBP
etaC-MI
o —•••• ,
NBP LkUp =:LaserWriter(g)*
Chapter Eight Macintosh Transport: AppleTalk
NVE
7;
=: Laser Wrlter(o)*
Deb's Dreamy LaserWriter :LaserWriter(2)*
When a LaserWriter responds to this request, it generates a NBP
Response such as Deb ' s Dreamy LaserWriter: LaserWriter®*,
which is shown in figure 8.41.
Deb's Dreamy LaserWriter
:LaserWriter@*
NBP Response
Figure 8.39
IhelaseiWritei inspects
llieNVianilpfoceedsio
Ipllillilie NBP lookup
Bequest.
Figure 8.40
IhelaserWfiter
completes the NVE by
adding its name to the
NVE.
Figure 8.41
IhelaserWiiter puts the
completed NVE into an
NBP Besponse packet.
Part Two Macintosh Netv\'orking
Figure 8.42
IlieLaseiWriierpuisilie
completed NBP Response
inioaODPDaiagiam. Hie
daiagiam is addressed to
the sender of the original
message.
The responding devices essentially fill in the blanks of their respec-
tive NVEs. Then, the process reverses order. After a LaserWriter
responds, its NBP Response is placed into a DDP datagram. The
“From" address is simply the address of the LaserWriter, which it
already knows; the “To” address is referenced from the source
address of the incoming request previously received. The complete
message looks like the one shown in figure 8.42.
NBP Resp
Deb's Dreamy
LaserWriter:LaserWiiter@*
DDP
From: 12.33.144
To: 12.255.2
Thus, the response is:
F rom : 12.144.133 (the LaserWriter)
To: 12.33. 144 (the Macintosh)
NBP Resp
Deb's Dreamy LaserWriter: LaserWriter®*
Again, let’s translate this response into English.
“OK, I hear you, Macintosh! I’m a LaserWriter in your current zone:
my name is “Deb’s Dreamy LaserWriter” and my address is
12.144.133. Since you told me what address you are when you sent
your NBP Lookup Request, I’ll send this information to you right
away.”
Chapter Eight Macintosh Transport: AppleTalk
Of course, multiple LaserWriters can respond to the NBP Lookup
Request, and as they do, your Macintosh displays their names in
the Chooser (see figure 8.43). Then, when you select a printer, the
Macintosh simply remembers the name of the currendy selected
printer. This name is then stored in the memory of the Macintosh
This is why, for example, when someone moves their PowerBook
from the office network to their home network, the office printer
will still be selected. To solve this problem, the LaserWriter must
be reselected from the Chooser.
Figure 8.43
/UierperiDtmiaoasiiiiilai
process 10 ihe
laserWtiier.ilieMac
processes die UAPsid
NBP anil ilieo places die
name ol die laserWriier
(and any Older lesponding
lWs)inioilieMaciniosli
Chooser.
This process also offers an explanation for the rare event that
occasionally happens when someone selects a printer on Friday,
then the network is altered over the weekend and the printer just
so happens to choose a new node number for itself. WTien Monday
comes, the user receives a “...can’t locate printer" message. Again,
reselecting the printer from the Chooser solves the problem.
This NBP request /response conversation happens every time a
Macintosh user selects a ser\ace from the Chooser. Because the
NBP delivery mechanism involves broadcasting (or multicasting),
NBP traffic can be a significant part of AppleTalk network traffic.
Part Two Macintosh Networking
AppleTalk Routing
Big AppleTalk internetworks are created by connecting little
AppleTalk networks. Routers are the devices that are used to
connect the networks. AppleTalk routers are an important part of
many AppleTalk networks. In addition to providing a mechanism
for growth, they also are used to provide traffic isolation and a
means for logically grouping (organizing) network services. In this
section, we’ll explore how AppleTalk routers accomplish these
basic tasks.
Routing Tables
As discussed earlier, AppleTalk routers are used to physically and
logically connect network segments. Each AppleTalk network that
connects to a port on the router is assigned a number, or a range of
numbers, that identify that particular network. These AppleTalk
network numbers are key to the operation of the router.
AppleTalk routers rely on tables, stored within the router, to
fonvard AppleTalk datagrams from one netw'ork to another. The
routing tables keep track of all networks by containing an entry for
each network number. For each network number on the internet,
the routing table includes the distance of each network (measured
in hops, which is the number of routers between the router and the
destination network), which port on the router should be used to
connect the destination network, and the AppleTalk node ID of the
next router. This is shown in Table 8.2.
Table 8.2 A simple routing table.
Net#
Distance
Port
Next Router #
10
0
1
0
20
0
2
0
Chapter Eight Macintosh Transport: AppleTalk
Net#
Distance
Port
Next Router^
30
1
3
12
40
2
3
12
RTMP, or the Routing Table Maintenance Protocol, was Apple’s
only protocol that maintained routing tables among the routers of
an AppleTalk internet. Witli RTMP, these routing tables are regu-
larly updated every 10 seconds. This is accomplished by each
router exchanging routing tables with the other routers on the
network. When a router receives a new routing table, it compares it
to the existing table. If a new network has been added, or a network
distance has been changed, the router updates its table.
RTMP traffic is normally present only between the routers on the
internet, but still represents a certain percentage of the total traffic
on the network. One of the problems with RTMP is the regular
transmission of the routing tables, which occurs even when the
network is stable and the network numbers and the routing tables
remain unchanged.
To better understand the function of RTMP, let's return to the
classroom analogy. As shown in figure 8.44, when the students
pass notes between themselves, they frequently have to use an
intermediate student as a router. Let’s consider that a student is a
two-port router (her arms represent network ports). The students
who are acting as routers need to know the names of other stu-
dents and the corresponding student-routers to which they are
connected. These student-routers also need to know which hand
(port) to use.
Part Two Macintosh Networking
Figure 8.44
The siudenis in the class
pass their messages
thiough routers.
Figure 8.45
Each student router, Pat
and Paula, has her own
routing inlormalion that
she uses in route the
other students'
messages.
To solve this problem, each student-router creates a special kind of
note that lists all tlie note recipients, the number of intervening
student-routers, which arm to use, and the name of the next
student-router in the note-passing chain. Then, once all the
student-routers have their own list, they pass it to the other stu-
dent-routers, who check the notes for updates. These routing lists
are shown in figure 8.45. This process continues at regular intervals
regardless of classroom changes.
Of course, under certain conditions, the note passing between the
student-routers can become a problem. The same is true in the
real AppleTalk world, as this continual RTMP traffic can become a
burden on large networks with many routers (particularly when
the routers are connected to WAN links of limited bandwidth).
Chapter Eight Macintosh Transport: AppleTalk
This routing technique, which uses a list of destination networks
with their corresponding distances and the next router in the
chain, is called vector-distance routing, or Bellman-Ford routing. It
is a simple routing algorithm (as illustrated in figure 8.46), that
attempts to find the shortest path for a datagram by minimizing
the number of hops. Other protocols, such as DRCnet and TCP/IP,
use other routing algorithms that may be more efficient under
certain circumstances. Apple is currently investigating other
routing algorithms for possible future adoption and inclusion into
AppleTalk.
Figure 8.46
Apiilelalk datagrams aie
routed by aiiempiing to
minimuethenumbeiol
hops. In this eiample,
dalagtams travel directly
Itomfll-20iod[il40.
The problem with vector-distance routing is that minimizing the
number of hops doesn’t always route the datagram through the
quicker path. Consider the example shown in figure 8.47. Network
#61-70 is one hop away from Network #11-20, but this connection
is made through a relatively slow 9600 baud network link. If we can
also reach Network #61-70 by making three hops over Ethernet
connections, the throughput will be dramatically improved. With
our student analogy, it might be faster to pass a note through the
three hyperactive students who eat chocolate all day instead of
waiting for one sleepy-eyed student to act.
Part Two Macintosh Networking
Rgure 8.47
AppleTalk datagiams
mightgolasieiDvera
louie that has more hops.
Iherelore.amcihoilo[
ariilicialhopadjusimeiii
isolieorequireii.
AppleTalk datagrams might go faster over a route that has more
hops. Therefore, a method of artificial hop adjustment is often
required.
Some AppleTalk routers permit the artificial adjustment of a
datagram’s hop count. Instead of the datagram’s hop count being
increased by one when going through the router, it could be
increased by two, which could force network traffic to take another
less cosdy, more efficient path. Apple’s new Apple Internet Router
(AIR) also adjusts liop counts in large networks so as not to exceed
the limit of 15 hops (see figure 8.48).
Remember that a key aspect of this routing technique is that the
routers are not aware of the entire route for any given datagram
destination. All that any given router knows for sure is the next
router in the chain.
Chapter Eight Macintosh Transport: AppleTalk
houioi s<tiim
Router Nome: Rpple Internet Router
p ,0> Pnr.UrPcrt
' Mo4*«t«Por!
DUUD
$ LootITtfc
CtN*rr«t
S3: Elh«rT«k
^ y TuM>ti
Ctt«rrt*t
AC«r*
Actrt*
C*n*»g I-555-»234 yU ACME 96 _
N«(: 1000, Zortr: OMton
Nons««4
3S 190 4430,80.13024 1
i Router Setup Port Info I
P>i«sta«1 P*rt: ^ Mo4*mPorl
A«««ss $ Lae»rr4k
P»rt 0»seri|itto*:
P«rt SUtus: (i) Aetw* Q
( Hide... ~]
P*rt: ® S*»i O honnyi
^ 1«”0 I
: [ Bwto*'
[Get Zones... ]
h:
Sttttottcs t«rt r»Mt 1 /1 2/93 3 ;3S Mi
s?
SUtistic* T*t4l E(li*ri»«t Prtatrr P*rl
RTMP versus AURP
To help solve the problem of excessive RTMP traffic, particularly
on WANs, Apple has developed a new routing protocol called the
Apple Update Routing Protocol, or AURP. AURP updates the
routing tables only when a change has been made to the network.
(Typically, this means whenever a new network has been added
to the internet.) With AURP, our student-routers would only
exchange routing notes when new students are added to the
Figure 8.48
Sample stieens from the
Apple InierneiHopiei.
Part Two Macintosh Networking
classroom. AURP is not intended to replace RTMP, which remains
a viable protocol for small- and medium-sized LANs; rather, AURP
is seen as a complement to RTMP.
.\pple describes RTMP as a “no news is bad news” type of routing
protocol. This means that if the RTMP updates aren’t seen every
ten seconds, then something’s wrong. Instead, AURP is categorized
as a "no news is good news" protocol, since as long as things are
quiet, the routing information must be okay. The AURP protocol
also defines a method of tunneling, or encapsulating, AppleTalk
datagrams within other netw'orking protocols. (The idea behind
protocol encapsulation will be covered in the next section.)
The first product to support AURP is the Apple Internet Router
(AIR). This new product supersedes the AppleTalk Internet Router
2.0. AIR, like its predecessor, runs on a suitably-configured
Macintosh. System 7 is required, along with a minimum of 4 MB of
RAM. PowerBooks are not recommended. AIR fully supports AURP
over LAN and WAN connections using a modular approach. AIR
also includes support for the industry standard Simple Netw'ork
Management Protocol (SNMP) which provides for remote manage-
ment.
The standard network supported by AIR is dial-up access over
standard phone lines (see figure 8.49). For a dial-up connection,
Apple recommends a minimum of a pair of V.32/9600 baud
modems. Obviously, the faster the modem, the faster and more
responsive the AppleTalk connection will be. Since AIR was
developed in a modular fashion, additional capabilities can be
simply plugged in as needed. Apple offers several optional prod-
ucts including an AppleTalk/X.25 and AppleTalk/IP extension.
Chapter Eight Macintosh Transport: AppleTalk
Figure 8.49
A MCI diagram ol the
Apple Internet Hauler
working over
The AppleTalk/X.25 Wide Area Extension, shown in figure 8.50,
enables multiple AppleTalk networks to communicate through an
X.25 wide area network. Apple’s Serial NuBus card is required to
connect to the X.25 service. X.25 will be covered later on; briefly,
X.25 is a wide-area networking ser\ace that is offered by service
providers world-wide. These providers (such as Tymnet and
Telenet) can offer customers local phone numbers that are used to
connect to the system. One chief advantage of X.25 connections is
that, unlike conventional phone lines which are billed solely based
on time regardless of traffic, X.25 uses traffic as a prime determin-
ing factor in billing. Considering that most network traffic is not
continual and tends to be “bursty,” X.25 is very often a cost-
effective alternative to traditional dial-up lines. A familiar example
of X.25 networking can be found at most banks, since many ATM
machines use X.25 links to provide wide-area access to your
financial data.
Madntoih
SlKVlCtt
Formats
ApplaTalk
Macintosh Running Apple Inlamat Router
_
Macintosh Running Apple mtamal Router
r App.aT.ik y/ x.25 )
r x.25 H )
5 Loc.rr.m ? ^ rs-2m,422 ?
5 RS-232 422 / S LocelTalk ?
Macintoalt
Sanricat
Figure 8.50
A HeiPICI diagram ol Hie
Apple InieiaelRpuief
using X.25 nelweiking.
LocalTilk
LoeatTalK
Part Two Macintosh Networking
The other optional AIR module is the AppleTalk/IP Wide Area
Extension. Diagrammed in figure 8.51, it links multiple AppleTalk
networks over a TCP/IP network. The AppleTalk/IP extension
is supported on Ethernet or Token Ring cabling. As with the
X.25 option, the IP option relies on a networking trick known as
tunneling.
Rgure 8.51
ANeiPICIdiagiamofilie
Apple InierneiRouler
usipgIPlunneling.
IP-Only
Ethernet
Macknoth kladnloah
Protocol Encapsulation and Tunneling
Using figure 8.52 as an example, imagine that you want to send a
letter to your uncle, who lives on a mountain across the country, as
quickly as possible. You could send the letter with an overnight
carrier, but for some reason, the carrier won’t deliver the letter up
the mountain. The regular mail service does go up the mountain:
unfortunately, sending it this way would take two weeks for
delivery. You then remember that you have a cousin that lives in
the town at the base of the mountain who can receive the over-
night delivery service. So you proceed to address the letter to your
uncle and affix the required postage for regular mail delivery.
Then, you place die addressed letter inside the overnight delivery
pouch and address it to your cousin. You don’t want to make your
cousin go up die mountain, so you place a note inside the over-
night pouch instructing him to simply drop the enclosed letter into
the nearest mailbox. This way, the local mail will deliver the letter
within a day or so. Thus, the total delivery time is only two or three
days.
Chapter Eight Macintosh Transport: AppleTalk
This scenario can be directly applied to networking as well. The
process of placing one transport protocol inside another is called
protocol encapsulation or tunneling. Tunneling AppleTalk inside of
another protocol, such as TCP/IP or DECnet, might be necessary
or desirable for several reasons.
One reason could be that an organization's wide-area network only
supports a certain protocol. This has been fairly common in the
past because a number of routers have only supported a single
protocol. For example, many companies that have extensive wide-
area DECnet networks interconnect them with DECnet-specific
routers. For these companies to be able to offer AppleTalk services
over the network, they would have to scrap all their existing
DECnet routers and replace them with multiprotocol AppleTalk/
DECnet routers.
Part Two Macintosh Networking
Figure 8.53
AKeiRCIiliapiiiofllie
Applelalk/DECnellunnel.
Ip||iiseiiapipl8,eaclil/A)l
haslwoEUierpet
coniiollers.Ilieiuppel
acis as ap AppleTalk
fpuier.sppetwofk
pupibeisipusibeppique.
Alternatively, they can tunnel AppleTalk protocols inside tlie
DECnet protocol. In this case, an AppleTalk datagram that is
directed to a distant network is wrapped inside a DECnet packet by
a special device and routed over the wide-area DECnet network to
another special device, where its AppleTalk datagram is then
extracted from its DECnet encapsulation and then passed along
using AppleTalk protocols to its final destination.
In the case of AppleTalk/ DECnet tunneling, the special device
happens to be a DEC VAX that’s running the AppleTalk for VMS
and DECnet protocols simultaneously (see figure 8.53). The
AppleTalk for VMS software establishes a connection with the
DECnet software and performs the encapsulation and
decapsulation of the AppleTalk datagrams.
n
■ri
-m
AppleTalk Net #10-20
SCT3
DECnet-only
Ethernet
AppleTalk Net #21-30
Macintosh
As mentioned before, with tlie new Apple Internet Router, tunnel-
ing is also possible with X.25 and TCP/IP networks. The concept
remains the same — AppleTalk datagrams are placed inside the
foreign protocol and then routed accordingly to the corresponding
Apple Internet Router (AIR) over the wide-area connection.
Routers that link an AppleTalk network to the foreign protocol
network are called exterior routers. Routers that perform
AppleTalk-to-AppleTalk routing within the internetwork are called
interior routers.
Chapter Eight Macintosh Transport: AppleTalk
When the AIR (or any AURP-compliant router) is used in a tunnel
configuration, the tunnel side of the connection of the exterior
routers appears as an end node of the encapsulating protocol.
Using IP encapsulation as an example, when two AURP-compliant
routers communicate over a TCP/IP link, they appear as two native
TCP/IP devices. On the interior side of the connection, however,
the routers must appear as Phase 2 routers that speak RTMP to the
other interior routers.
Wlienever an AppleTalk datagram is encapsulated in a foreign
protocol, the AURP-compliant router adds a special AURP header
in addition to the addressing header of the foreign protocol. This
special variable length AURP header is called a domain header or
DI. Technically, the AURP domain header extends the standard
32-bit AppleTalk address (network, node and socket) and creates
a complete AURP wade area address.
Before AURP, one of the key rules of AppleTalk network configura-
tion was the uniqueness of AppleTalk network numbers. With the
addition of the AURP DI number, it is no longer a requirement that
the network number must be globally unique (although it’s still a
good idea). No\\f, with the addition of the DI the complete address
is domain ( network . node . socket ) .
Adding the DI is similar to adding an area code to a phone number.
For example, the phone number 555-1212 is certainly not unique
in the whole US phone system. But, by adding a unique three-digit
area code to the beginning of the number, it becomes a valid
directory assistance number for any area code.
To better illustrate this point, let’s use an example of two uncon-
nected LANs. Figure 8.54 shows one LAN in Philadelphia having a
Phase 2 network range of 10 to 20. The other LAN in San Francisco
has a range of 15 to 50. Under the pre-AURP Phase 2 rules, these
two networks could not be linked together because their network
numbers overlapped. Both LANs have a network number 15, and
o
Part Two Macintosh Networking
ngure 8.54
With IlieAlIRP Domains,
AppleTalk network
numbers do not
necessarily have to be
uoique. In this example,
there are two devices
osing the 15.22.129
address. The addition el
the DURP Domain makes
them unique.
therefore a conflict would occur. With AURP Dls, each LAN could
have a unique domain number. Philadelphia’s domain number
could be 1 and San Francisco’s domain could be 2. This would
create unique addresses for both sides. 1(15.22.129) is now differ-
ent from 2(15.22.129). Of course, the Dls must be unique on the
tunnel. If the tunnel is a TCP/IP tunnel, one sure way to insure that
the Dls are unique is to use TCP/IP addresses as the Dls.
2(15.22.129) 1(15.22.129)
Zone
Domain: 2
Net #15-20
San Francisco
Domain: 1
Net #10-20
Zone: Phiiadelphia
Conclusion
The Macintosh supports a number of transport protocols. Its
native protocol, AppleTalk, provides plug-and-play connectivity.
AppleTalk nodes use dynamic node addressing and name binding
to self-configure, which avoids the manual configuration required
by other protocols such as DECnet and TCP/IP. Since its inception,
Apple has continued to evolve AppleTalk, the most recent changes
providing enhanced routing over wide-area networks.
Macintosh
Media/Cabling
ne of the more visible aspects of a Macintosh
network is the medium used to make the
connection. This chapter covers the various
connection media, such as LocalTalk and
Ethernet. It covers the advantages and disad-
vantages of each medium, giving you the
knowledge to choose the right medium — or
combination of mediums — for your network.
LocalTalk/Phone-Type Connectors
LocalTalk, along with its phone-type variants, is one reason for the
popularity of Macintosh networking. Its low cost and easy instal-
lation has set the standard for desktop networking. Although
LocalTalk is rapidly being supplanted by Ethernet, it still offers a
viable solution for many Macintosh users.
Part Two Macintosh Networking
Figure 9.1
Apple's locallalk
coppecioisysieip.
ICouriesif of Apple
Compulei.lpcj.
LocalTalk
LocalTalk (see figure 9.2) was the first network cabling system
available for tlie Macintosh. Introduced in 1985 along with the
LaserWriter printer, LocalTalk was a low-cost plug-and-play
solution in a world of thousand-dollar Ethernet cards. LocalTalk
and its variants are still used today, but its use is rapidly declining.
LocalTalk provides a bandwidth of 230.4 Kbps, which is pretty
quick compared to a 9600 baud modem connection. Compared to
a 10 Mbps Ethernet connection or a 100 Mbps FDDl connection,
however, it’s pretty slow.
The LocalTalk cabling system normally connects to a Mac’s Printer
serial port (shov\Ti in figure 9.3), and to the appropriate LocalTalk
ports of other devices.
NOTE: The Modem port is not used for LocalTalk connections unless you
intend to run router software (such as the Apple Internet Router) on your Mac
to route between two LocalTalk segments.
0
Chapter Nine Macintosh Media/Cabling
LocalTalk
Cabling Layer
Figure 9.2
locallalk resides ai Die
Cabling layer of Uie
NetPICI.
Modem Port Printer Port
The heart of LocalTalk is the small connector box which contains a
small transformer that electrically isolates the network connection.
It has three connections. One side of the box has a length of v\dre
and a connector (either a circular 8-pin D1N8 connector or a D-
shaped 9-pin DB9 connector) that is used to connect to the
AppleTalk node. The other side of the LocalTalk box has two
receptacles that are used to connect the node to the chain of other
LocalTalk devices.
Figure 9.3
Bolb the Modem and
Primer ports are serial
ports; however, only ihe
Printer port can be osed
fotlocallallr.
Part Two Macintosh Networking
Apple’s LocalTalk connectors are rarely used today for several
reasons. First, when a LocalTalk network is indicated, it usually
makes sense to consider the LocalTalk-compatible phone-type
connectors. Farallon's PhoneNET, which uses twisted-pair tele-
phone-type wiring, is one such alternative. Another reason is that
the LocalTalk connectors do not have a positive locking arrange-
ment — they can be easily pulled out or disrupted. People have
been known to wrap electrical tape around the boxes or even Crazy
Glue the connectors in place to avoid this problem. Perhaps the
most important reason for the decline and fall of LocalTalk is that
it’s rapidly being replaced by Ethernet. The price of Ethernet
connections has dropped dramatically over the past few years, so
the cost differential is not as great as it was in the past.
LocalTalk does have cable shielding, so in electrically noisy areas,
LocalTalk might be better titan the unshielded twisted-pair wiring
used by the phone-type connectors. For most small installations
where the anticipated network traffic is modest, you’ll want to
consider the phone-type devices.
PhoneNET and Other Phone-Type Connectors
Farallon was the first company to offer a functional replacement
for LocalTalk. They replaced the DIN8 connectors of LocalTalk
with the positive locking RJ-11 connectors found on most tele-
phones (see figure 9.4). They also replaced the expensive shielded
LocalTalk cable with conventional twisted-pair phone wiring. The
PhoneNET connectors were, and remain, completely compadble
with the LocalTalk Link Access Protocol (LLAP), so switching over
to PhoneNET from LocalTalk requires no software changes or
special configuration.
Chapter Nine Macintosh Media/Cabling
The success of PhoneNET and similar products was twofold. Small
companies were able to create a simple, inexpensive LANs in an
hour or so by simply going to Radio Shack and buying a spool of
phone wire, a box of RJ-1 1 connectors, and a $15 crimping tool.
Large companies soon found that it was possible to integrate the
PhoneNET connectors into their existing wiring schemes — wiring
schemes previously used for connecting “dumb" terminals to
mainframes and minicomputers over twisted-pair RS-232 wiring.
The PhoneNET connectors also made it possible to move to a star
topology (or radial topoIogjO. instead of the daisy-chain topology
of LocalTalk. PhoneNET currently supports two kinds of stars:
passive stars, in which each of the segments is interconnected at a
panel or junction block, and active stars, where the segments join
at a LocalTalk repeater. Farallon’s first repeater was called the
StarController.
Hgure 9.4
TtieFaiallonPhoneNil
Conneciofuses
lelephonettifleRJ'll
locking conneciois.
ICouiiesyofFaiallon
CompuiiogJ
Part Two Macintosh Networking
Since the star topology only requires a single connection at the
node, only one RJ-l 1 receptacle is required. When a PhoneNET
connector is used, the extra receptacle is filled with a terminating
resistor. Other brands are self-terminating and don’t require a
separate resistor. Farallon also offers a single-receptacle connector
called the StarConnector. This small connector (see figure 9.5),
plugs directly into the printer port of a Macintosh, and is ideally
suited for star networks. StarConnectors are also useful in pairs,
where they can be used to connect two devices (a PowerBook and a
desktop Mac, for example) with a single R]-l 1 cable.
Figure 9.5
IheFatallonHioneNlI
SiatCmecioipluos
diiecilyiniDihePiinier
port Used in star
neiworks, die device is
self-ieiminaling.
(CounesyolFaiallon
CompuliRp.)
Today, there are many LocalTalk-compatible products. In addition
to Farallon, there are several companies that make the phone-type
connectors and products.
Design Considerations
If your network consists of a dozen or so nodes in a centralized
location, and your bandwidth requirements are modest, then a
LocalTalk or PhoneNET daisy-chain might make sense. With either
device, you’ll be able to add nodes to the ends of the chain, but if
you need to add an additional connection somewhere in the
middle, the entire network will be disrupted.
Chapter Nine Macintosh Media/ Cabling
A more flexible solution for small workgroups is to use a bus
topology. With this system, a single cable is used as a backbone.
Each phone-type connector then plugs into the bus. Connections
can be made and broken at any time without disruption to the
network. This is only possible with the phone-type connectors,
such as PhoneNET, and will not work v\qth the LocalTalk
connectors.
The next step up is to go with a star topology. The choice of a
passive star can be limiting, with restriction on the number of
devices and distances. In most cases, passive stars are problematic
and not worth the trouble. The other option is to go with an active
star. The active star involves the use of a LocalTalk multiport
repeater. Products such as the Farallon StarController (shown in
figure 9.6) and the Focus TurboStar avoid the problems found in
the passive star topologies. These repeaters are a good choice for
the LocalTalk network that is spread out through a large building.
They’re even a better choice if you’re able to use existing twisted-
pair wiring.
Figure 9.6
The Faiallon PhoneNET
SiaiConirolletisan
example ol a niuitipDii
locallalluepeaier.
(Copriesyol Farallon
Cooipoling.)
Part Two Macintosh Networking
Planning Ahead
If you're going to go the tv/isted-pair phone-type connector route, be sure to
use twisted-pair wiring that's capable of supporting twisted-pair Ethernet.
Phone-type connectors require a single twisted pair, while twisted-pair
Ethernet requires two twisted pairs. LocalTalk signaling will work over the
Ethernet twisted-pair wiring, but not vice-versa. So plan for the future and wire
for Ethernet. The additional cost of the cable should be negligible compared to
the installation cost. Just make sure you use the appropriate twisted-pair
wiring, as there are different kinds of wiring for different applications.
As you consider a LocalTalk or phone-tvTJe connector network,
keep in mind that the additional cost of die LocalTalk hub in-
creases the per-device cost. Now that Ethernet connections are so
inexpensive, you may want to consider spending just a bit more
to go Ethernet. If all your de\ices — present and planned — are
equipped with LocalTalk, then choosing it as a cabling system may
not be a bad choice. But if you plan to add other PCs, workstations,
or printers that don’t offer LocalTalk options, maybe it’s better to
opt for Ethernet at the outset.
Wlien Ethernet is chosen as a backbone cabling system, it used to
be that the only option for connecting LocalTalk-only devices
(such as LaserWriters) was to install a LocalTalk-EtherTalk router.
This is an expensive cure when you may only have a limited
number of LocalTalk devices to connect. It also adds the adminis-
trative overhead and performance concerns of adding another
router to the network. Several years ago Dayna Communication
introduced a device called EtherPrint. It acted as a LocalTalk-to-
Ethernet bridge. It essentially adapted the LocalTalk device to the
Ethernet network. Today, these devices, diagrammed in Figure 9.7,
have gained in popularity and functionality.
Chapter Nine Macintosh Media/Cabling
Figure 9.7
Seveial companies
piovidelocalialklo-
Eiherlallditiiloes
ladapiefslitiatcanlie
used IP connect
locallalkonly devices,
sucliaslaserWriiersand
PowefBooks,toihe
Eilieinei
Companies such as, Asante (AsantePrint), Compatible Systems
(EtherWrite), Dayna (EtherPrint Plus), Farallon (LocalPath), Sonic
Systems (SuperBridge) and Digital Products (SprintTALK) make
LocalTalk-Ethernet bridges. The new generation of devices is able
to connect multiple LocalTalk devices — Macs and PowerBooks
included. The Farallon and Sonic Systems offerings are software
solutions that run on a Mac with the appropriate network inter-
faces. The Farallon product can even connect devices to Token
Ring networks. Many of these devices support security features
that can limit access to the LocalTalk device.
Ethernet/EtherTalk
At one time, Ethernet was to cabling systems what FDDI is today —
it was expensive and offered a bandwidth well beyond the require-
ments of most applications. Today, Ethernet is fast becoming the
modern equivalent of RS-232. Nearly all computer systems offer
Ethernet connections. With the advent of very large-scale
intergration, the Ethernet components have been reduced to a
single chip, wiiich has driven the cost of Ethernet cards down to
Part Two Macintosh Networking
less than $200. Fortunately, these price reductions aren’t limited to
the PC world, as Macintosh Ethernet cards are fast becoming a hot
commodity.
Ethernet Basics
Even though Apple developed LocalTalk to provide a basic physical
connection between devices, Apple also recognized the need to
provide alternative wiring choices to their customers. One of the
most popular local area networks (LANs) is Ethernet (see figure
9.8). Originally developed by Xerox in collaboration with Digital
and Intel, Ethernet is used by many computer vendors as a wiring
media for networking. Companies such as Digital and Sun use
Ethernet to run a variety of networking protocols. In fact, Ethernet
was developed with multiprotocol support in mind. A single
Ethernet network can support many different protocols at the
same time.
Figure 9.8
Elhernet resides at (lie
Cabling layer of Uie
NeiPICI.
Ethernet
Cabling Layer
When AppleTalk protocols run over Ethernet cables, Apple calls
this EtherTalk. Whereas LocalTalk cables have a bandwidth of
230.4 Kbps, Ethernet has a bandwidth of 10 Mbps. So instead of
being limited to 32 nodes, as with LocalTalk, EtherTalk networks
can support thousands of devices. Theoretically, with the latest
version of AppleTalk (Phase 2), a network can have over 16 million
Chapter Nine Macintosh Media/Cabling
devices. Of course, on a single cable you would run out of room to
connect all those devices, but Ethernet cables are often “con-
nected” by network bridges, microwave links, and even satellites to
other Ethernet networks to create an “extended” Ethernet LAN.
Many large companies have extensive world-wide Ethernet LANs
with thousands of computers produced by different companies.
The throughput of an Ethernet network is higher than LocalTalk.
Actual transmission rates will depend on many factors, such as
network traffic, size of the transmitted file, and performance of the
individual Ethernet controller. On average, you can expect an
Etliernet network to perform three to five times better than a
LocalTalk network. \Vliy such a difference? The factors limiting
Ethernet throughput are numerous and complex, but the speed of
the Macintosh CPU, hard disk, and Ethernet hardware, coupled
with network configuration, application performance, and other
network traffic all play a role in Ethernet performance.
All Macs and LaserWriters come standard with the hardware to
support LocalTalk communications. Some Macs (such as the
Quadra family) and LaserWriters (such as the LaserWriter IIG and
LaserWriter Pro 630) come equipped with built-in Ethernet hard-
ware. For those Macs that don’t have Ethernet, connections are
made with the addition of a networking card. LocalTalk
LaserWriters, such as the LaserWriter IINT, can connect to
Ethernet with adapter devices, such as Dajma’s EtherPrint de\ice.
Ethernet cards for Macs with card slots are made by Apple and
other vendors (see figure 9.9). The cost varies between $200 and
$400. Several companies also sell SCSI/Ethernet adapters for those
Macs, such as the Classic and the PowerBook family, without card
slots. These devices connect to the SCSI port of the Macintosh, just
like any other SCSI device, and dien connect to the Ethernet
network.
Part Two Macintosh Networking
Figure 9.9
A sampling olEiliernet
cards. IlieiwO'Piece
anils are for lire
MacinlPSliSEandSE/3D.
lire one-piece pail is lor
anyliuBus-eguipped
MacinlosIr.lCouiiesyol
Farallon Computing.)
ELAP software drivers are included with both the Ethernet cards
and the SCSI/Ethernet devices. This software provides new net-
work driver programs that give you the option, through the Control
Panel, to choose between LocalTalk and EtherTalk. Unless you
turn your Mac into a router, the AppleTalk traffic can only go
through one port at a time.
\
There are several variants of Etliernet cabling. Even though the
cabling is different, the electrical signaling remains the same.
Because of this, all Ethernet cable variations use the same ELAP
drivers. The only significant difference is the cable type and the
connectors.
Thickwire 10Base-5
Thickwire Ethernet (see figure 9. 10), is a stiff coaxial (one wire
inside another wire) cable about 3/8" in diameter; it employs a
15-pin D-style connector. The cable is terminated at both ends
with special resistive fittings that minimize signal reflections that
Chapter Nine Macintosh Media/Cabling
would otherwise degrade communications. Usually thickwire
Ethernet is employed as a central "backbone" running throughout
a building, although fiber-optic Ethernet is rapidly replacing
thickwire as a backbone media. Thickwire Ethernet permits a
maximum of 200 devices on a 1,640 foot segment. Thickwire
Ethernet is often referred to as lOBase-5 wiring.
Terminator Barrel Connector Terminator
Thickwire connections are established by clamping a device called
a transceiverto the cable. Most transceivers are installed by drilling
a small hole in the cable with a special tool. This is followed by
clamping the transceiver, which pierces the cable with sharp
contact pins that make electrical contact. This method of connec-
tion is sometimes referred to as a "vampire tap.” These taps can
only be made at regular interv'als along the cable. Most cables have
indicator markings every 2.5 meters to help position die transceiv-
ers.
Adding transceivers to a backbone cable is not difficult, but this is
not a cost-effective way to make a single network connecUon.
Usually transceivers are used to connect hubs (or repeaters), which
support the connection of multiple devices through a single
transceiver connection to the backbone.
Figure 9.10
Thickwire |10flase-5|
Elherneicomponenis.
(i.e. Computer, Hub, Router)
Part Two Macintosh Nehvorking
In the lOBase-5 nomenclature, the 10 refers to the bandwidth. All
Ethernet implementations have a 10 megabit bandwidth. The term
“Base” refers to baseband (as opposed to broadband). Baseband
means that the cable only supports a single communications
channel. The value of 5 refers to the maximum length of the cable
segment, which for lOBase-5 is 500 meters. There is a specification
for broadband Ethernet called lOBroad-36. It runs over a coaxial
cable and has a maximum segment length of 3,800 meters.
Figure 9.11
Iliinwjie(IOBase'2|
[Uieineicomiionenis.
Thinwire 10Base-2
Thinwire Ethernet (see figure 9.1 1) is thinner (about3/16 ") and
considerably more flexible than the original thickwire. It used to be
a popular choice to connect desktop devices and workstations.
Today, however, thinwire is being rapidly replaced by twisted-pair
Ethernet for desktop connections.
Terminator Tee Coaxial Cable Terminator
Thinwire Ethernet uses BNC (twist and lock) type connections and
allows 30 devices per 656 foot segment, over a maximum nettvork
length of 3,281 feet. Thinvwe is often referred to as "Cheapernet,"
or its more formal name of lOBase-2. This designation is similar to
that of thickwire: the 2 indicates a maximum segment length of 200
meters (actually, the maximum segment length is 185 meters, but I
guess they didn’t want to call it lOBase-1.8).
Chapter Nine Macintosh Media/Cabling
Connections are made with a tee connector, similar to LocalTalk.
One branch of tlie tee connects to the network device, while the
other two connections are used to connect to the network. When
thinwire is connected in a daisy-chain, the free ends of the last tees
must be terminated with special resistive end caps. Adding more
devices to a thinwire daisy-chain disrupts the network because the
chain must be broken. It is possible to disconnect a device at the
attachment point without disrupting the network chain.
Thinwire can also be configured in a star topology using a thinwire
Ethernet repeater. Here, each tee at the end of each branch of the
star must be terminated. The use of thinwire star topologies is
rapidly declining, due to the arrival of the newer, more flexible
twisted-pair Ethernet.
Twisted Pair lOBase-T
Lately, another Ethernet variant has started to become popular.
Twisted-pair Ethernet (see figure 9.12) has been around for several
years, but during the past two years it has virtually dominated
the desktop. With this system, Ethernet can be implemented
on standard unshielded twisted-pair wiring. Thinwire Ethernet
requires two pairs of wires that meet certain industry
requirements.
Figure 9.12
IwisiedpaifllOBase-H
Ellieineicompoflenis.
Part Two Macintosh Networking
The choice of wire may depend on local building or electrical
codes, or on the recommendations of suppliers. It’s always best to
check with the various codes and suppliers for detailed cabling
specifications. As mentioned earlier, it’s always wise to cable for
thinwire Ethernet even if you’re planning to use LocalTalk. It also
may be prudent to consider the anticipated wiring standards for
the newer high-speed cabling systems such as FDDI (Fiber Data
Distributed Interface). There is currently work underway to imple-
ment FDDI over copper twisted-pair wiring and to develop a 100
Mbps version of Ethernet that runs over four twisted-pairs. Often,
if your wiring strategy covers the most stringent wiring scenario,
you’ll be able to design for future growth and enhancements.
Unlike thickwire or thinwire tliat can be connected in a bus or
daisy-chain, twisted-pair Ethernet requires the use of a hub. These
hubs come in a wide variety of prices and configurations with a
varying number of ports and extra features. Some hubs offer a
modular construction that make it easy to provide additional
connections as required.
All the Ethernet card vendors for die Macintosh offer twisted-pair
Ethernet versions, with many cards offering multiple connectors
(thick, tliin and twisted-pair) on one card. Twisted-pair Ethernet
underwent some changes during its early years, and there have
been several implementations, but now the standard is set and is
widely known as lOBase-T.
Apple s Ethernet Cabling System
In January of 1990, Apple announced a new line of low-cost
Ethernet cards. The two new cards (one for the Macs with NuBus
slots and one for the Macintosh LC) use a separate attachment
unit, known as an Apple AUI (Attachment Unit Unterface) that
attaches to either thickwire, thinwire, or twisted-pair Ethernet (see
figure 9.13). Resembling LocalTalk connectors, these connectors
attach to the Mac or LaserWriter with a new style of connector.
Chapter Nine Macintosh Media/Cabling
This permits Apple to use these new compact connectors on the
motherboards of all their new machines. The appropriate attach-
ment unit (either thick, thin, or twisted-pair) is then connected to
the device. This approach offers Apple a single, compact connector
for their new products, while still offering the flexibility of three
attachment options. Apple has provided the specifications for
Apple AUl, so these new connectors are also offered by third-party
suppliers. These new devices will bring the ease-of-installation and
low cost of LocalTalk to Ethernet networks.
Hgure 9.13
Apple's EihernelNuBus
canKandcompuiefsso
egaippedl can canned lo
Apple's lliick. in nr
iwisied-pair transceivers.
Design Considerations
As with any network design, the physical location is an important
determining factor. If all your AppleTalk nodes are in the same
room, then lOBase-2 thinwire easily can be daisy-chained among
the devices, and will most likely be the cheapest solution. The
Part Two Macintosh Networking
other option is to use lOBase-T twisted-pair Ethernet. If you
choose lOBase-T, you’ll need a hub to interconnect the devices. If
you have just a few devices to connect, there are a number of
vendors that offer small, affordable hubs with anywhere from three
to eight ports. The additional price of the hub may make this
approach more expensive than thinwire, but the cabling should be
slightly cheaper and you may be able to assemble your own cables
more easily than thinwire.
If your network is large and spans many rooms or floors, a back-
bone Ethernet is probably needed. With this approach, a lOBase-5
or fiber optic cable is strategically placed through the building.
Often, the cable runs between wiring closets, where nebvork
equipment such as hubs, routers, and gateways are located. These
wiring closets are often used for phone connections as well. (Some
networking vendors offer integrated networking equipment to
consolidate the management of voice and data connections.)
With most extensive EtherTalk networks, a major design consider-
ation is the routing and isolation of traffic. Deciding where to place
routers and how best to link EtherTalk LANs to wide-area connec-
tions is frequently an important concern.
Other Cabling Systems
LocalTalk and Ethernet are the most popular cabling choices for
the Macintosh, but the Mac also supports a wide range of other
industry standard cabling systems.
Token Ring
Token Ring networks (see figure 9.14) operate using a different
principal than Ethernet. As discussed earlier, Ethernet devices
listen to the cable before transmitting, whereas Token Ring devices
Chapter Nine Macintosh Media/Cabling
wait their turn until an electronic token comes their way. Because
of this fundamental difference, Token Ring networks enjoy certain
benefits over Ethernet netw'orks.
Rgure 9.14
Token Ring MCI.
First, when it comes to traffic. Token Ring networks are self-
limiting. Unlike Ethernet networks, which degrade when excessive
traffic causes collisions and retransmissions. Token Ring networks
simply reach their maximum throughput and then level out.
Another advantage of Token Ring is that a node is always guaran-
teed access to the cable within a finite period of time. Ethernet
nodes play a statistical game where access to the cable is not
guaranteed. This makes Token Ring networks appealing for time-
critical, real-time, and process control applications.
Most decisions to select Token Ring technology (other than FDDI)
are not made because of its technical advantage, but rather to
connect to the IBM environment, where Token Ring is a popular
choice. There are currently two implementations of Token Ring — a
4 Mbps version and a 16 Mbps version.
Token Ring cards are more expensive than their Ethernet counter-
parts. Prices range from $500 to $800 per card. Apple and several
third-party vendors offer NuBus Token Ring cards.
FDDI
Likely to succeed Ethernet, FDDI (Fiber Distributed Data Interface)
is an ANSI and ISO standard network based on dual fiber optic
Part Two Macintosh Networking
Figure 9.15
FDDIHelPICT.
rings (see figure 9.15). FDDI has a bandwidth (data throughput
rate) of 100 Mbps. This is 10 times the bandwadth of Ethernet. Just
as Apple offered EtherTalk and TokenTalk drivers for Ethernet and
Token Ring wiring systems, tliey have also developed FDDlTalk
drivers. The Apple drivers currently support AppleTalk Phase 2 and
MacTCP. FDDI networks can contain 1,000 nodes no more than 2
kilometers apart, for a total aggregate distance of 100 kilometers.
Figure 9.16
A FDDI card lor MuS’
e(|uip|iedMacs.(Couriesv
olCodenollIecMDgy
Corp.)
FDDI cards (see figure 9.16) are still a bit on the expensive side at
over $ 1 ,000 per card, but just as the cost of Ethernet cards dropped,
the price of FDDI cards will come down as well. FDDI cards are
currently offerred by several companies; Codenoll and Impulse
Technology are two examples. While FDDI is still rare on the
desktop, it’s becoming increasly prevalent as a backbone cabling
system.
Chapter Nine Macintosh Media/Cabling
Although FDDI is gaining in popularity, there are other upcoming
standards vying for acceptance. A proposed “CDDI” standard
would offer the performance of FDDI over less-costly copper
cabling. HP and AT&T are proposing an upgrade to the Ethernet
standard to achieve FDDI performance levels (100 Mbps) over
lOBase-T twisted-pair wiring. Instead of two twisted pairs, this
approach requires four twisted pairs.
Serial RS-232/422
Serial communications only recently have become a popular
cabling medium for AppleTalk. Starting v\ath Apple’s Remote
Access Protocol (ARAP), many Macintosh users are using serial
connections and modems to dial in to remote Macintosh comput-
ers. ARAP uses data compression and buffering techniques to coax
the most out of the relatively slow dial-up links (see figure 9.17).
Remote Macintosh
Rgure 9.17
Applelalkflemoie Access
NelPICI.
Part Two Macintosh Networking
ARAP uses die client/server model to make the remote connection.
A Macintosh running the client portion of Remote Access dials in
to a Remote Access server. With Apple’s software, the server is a
Macintosh. There are other servers from third-party vendors that
use dedicated hardware devices. These servers, such as Shiva’s
LANrover, connect to multiple dial-up lines and also make a
network connection to LocalTalk or Ethernet networks.
Figure 9.18
AflCNinieiPICI.
ARCNET
ARCNET is a cabling system that is popular on IBM PCs (see figure
9.18). It runs over twisted-pair or coaxial cabling at a bandwidth
of 2.5 Mbps. A Macintosh version of this cabling system is now'
available, along with software that provides the ARCNET data link
drivers (ARCNET Link Access Protocol). For more detailed infor-
mation on ARCNET, refer to Chapter 10, “Living in an Intel/DOS
World.”
Wireless
As computers get smaller and smaller, the cabling systems that
used to connect them also tie them down to the desktop. The
solution is to eliminate the cabiing. Wireless networks are a recent
development that does just that. There are several wireless tech-
nologies available for the Macintosh.
One option for a wireless network is to use Apple’s Remote Access
with a celluar phone/modem combination (see Figure 9.19). This
Chapter Nine Macintosh Media/Cabling
makes sense for wide-area network connections for a limited
number of devices. For LAN connectivity, wireless technology may
be useful in locations where conventional wiring is difficult or
impossible to run. Motorola, the leading manufacturer of cellular
telephones, has a Macintosh product called EMBARC which
provides a one-way wireless messaging service for remote Mac
users.
Remote Macintosh
Figure 9.19
CellulaiMoileiii/ARA
NelPiCI.
There are also options for LAN mediums such as LocalTalk and
Ethernet. Photonics makes LocalTalk devices that use reflected
infrared to link a number of nodes (see figure 9.20). The infrared
devices focus their energ>' at a single point on the ceiling.
Motorola has developed a wireless version of Ethernet called Altair
II (see figure 9.21). These devices use low-power radio waves as a
transmission medium. Altair’s transmission rate of 5.7 Mbps is
somewhat less than Ethernet bandwidth. Setup is easy. Each
Ethernet device connects to a small desktop send/ receive module.
Part Two Macintosh Networking
Figure 9.20
Monies makes a
locallalk device iliai
usesinfiaiedwavesasa
conneciien medium.
These desktop modules transmit radio waves to control modules
that connect to walls or cubicle partitions. The send/ receive
modules support all kinds of Ethernet adapters and cost around
$1,200. The control module can be used alone or connected to a
conventional Ethernet cable. These devices can handle up to 50
wireless devices.
Figure 9.21
Motorola's Allaiiil
loclinology uses radio
waves 10 carry Eihernei
iransmissions.
Compared to conventional wired networks, these new technolo-
gies are still somewhat expensive and are only cost-effective in
those cases where wiring is difficult or where reuiring costs would
exceed the cost of the wireless components. Expect wireless
communications to continue to increase in popularity as Apple’s
Newton technology and other handheld computers become
popular.
0
Chapter Nine Macintosh Media/Cabling
WAN Media
The cabling systems mentioned previously are typically used by
LANs. But when Macs need to be networked over longer distances,
other technologies are needed. However, compared to the cost and
convenience of LAN connections, WAN options are often limited
and expensive.
Conventional Dial-Up
Modem access was discussed earlier in the context of remote
access. A Macintosh user dials into a server on the LAN and con-
nects to services using a pair of modems. This is fine for the single
user, but what about connecting two (or more) LANs over a dial-up
connection?
There are modem-based products that work over conventional
analog dial-up lines that can be used to link multiple AppleTalk
LANs. An example of this can be found in Apple’s Internet Router
(AIR). In its standard configuration, as shown in figure 9.22, it
offers dial-up access over standard phone lines. Apple recom-
mends a pair of V.32/9600 baud modems as a minimum configura-
tion. Farallon, Shiva and other companies make similar products.
Figure 9.22
Ihe Apple Internet Rpiiter
can be used to link twn
temolelocallalk
neiwptks over a dial up
line.
The problem with these products is the relatively slow dial-up
connection. This type of connection is slow even for the single
Part Two Macintosh Networking
Remote Access user, and more so for a number of LAN users.
Conventional dial-up connections are also plagued by noise and
disconnections that are more likely to occur with lengthy connec-
tion times.
Figure 9.23
Engage's Syncnouier can
perloimApplelallaouiing
Dver Switched S6K lines.
Switched 56K
A faster and more reliable dial-up service is Switched 56K Service.
This is a digital 56 Kbps service provided by AT&T over their
ACCUNET system, by US Sprint, and others. Obviously this is
much faster than a conventional 9600 baud dial-up line: it’s also
much more reliable. This kind of service is very well suited to
connecting remote AppleTalk LANs. Engage Communication Inc.
offers an AppleTalk router (SyncRouter LTNT) that connects to
Switched 56K services. A NetPICT of the SyncRouter is shown in
figure 9.23.
ISDN
ISDN stands for Integrated Services Digital Network. The promise
of ISDN is high-speed, digital communications that provides
network, voice, and video to consumers. Think of ISDN as the
digital replacement technologj^ for your telephone. Unfortunately,
compared to other countries such as France, the United States is
lagging in the implementation of ISDN services. This is starting to
change as certain areas are now starting to get ISDN.
ISDN uses three channels: two 64 Kbps “B” channels and a single
16 Kbps “D” channel. The “B” channel carries voice and data, while
Chapter Nine Macintosh Media/Cabling
the “D" channel carries signalling data and can be used for lower-
speed transmissions. As ISDN services become more readily
available they should begin to displace conventional modems as
means of connecting AppleTalk services. Engage Communications
Inc. has just announced AppleTalk routers that use ISDN as a link
(see figure 9.24). They offer a LocalTalk version that provides 64
Kbps throughput and an EtherTalk version that uses both channels
and can achieve a throughput of 128 Kbps.
Figure 9.24
Engage's SyncRoolei can
peilorniApplelallt
fouling over ISDN lines.
Leased Services
Other high-speed services are leased for a specific period of time.
These services are incrementally based with T1 ser\ices offering a
bandwidth of 1.544 Mbps, T2 with 6.312 Mbps, and T3 with 44.736
Mbps. These high speeds come with high price tags as well, making
them suitable only for large organizations with demanding traffic
requirements.
riiese leased-line services are usually coupled with high-speed
multiprotocol routers from companies such as Cisco, Wellfleet,
DEC, and IBM. All these vendors provide AppleTalk routing over
leased lines. These multiprotocol routers cost anywhere between
S3, 000 and $20,000. They provide an>4hing from a single WAN and
Ethernet connection to units that have multiple WAN and LAN
connections over Ethernet, Token Ring, and FDDI. Figure 9.25
shows a NetPICT of a T1 -connected WAN.
Part Two Macintosh Netw'orking
Figure 9.25
Mannendofslsuchas
Cisco, Wellileei, DEC, anil
IBM) make routers that
support higlispeed II
leased'lineWAII
connections.
X.25 Packet Switched Services
Chances are that when you use your Automated Teller Machine
(ATM) you’re also using the services of a packet switched network
known as X.25. ATM machines enable customers to access their
checking and saving accounts anywhere in the world. Obviously,
each ATM machine doesn’t have everyone’s personal account
information contained within; instead, it accesses your account
information from a centralized location. Other examples of packet
switched networks are bulletin board sei'vices such as AppleLink
and CompuSei've. Users of these systems connect to the system
over local phone connections.
This is all made possible with X.25 networking. In an X.25 network,
services are locally provided over local phone lines. Once con-
nected, the service proxades a logical connection to anywhere on
the entire system.
In the past, packet switched networking was unsuited for
AppleTalk networking because of the overhead imposed by the
RTMP routing protocol. Now, with AURP, packet switching has
become a viable alternative for AppleTalk WANs. X.25 routing is
being offered as an optional extension to Apple’s Internet Router.
One primary advantage of an X.25 connection is that the costs are
primarily based on usage rather than connect time. In contrast,
with a conventional dial-up line, cost is mainly based on con-
nect time. The phone company doesn’t care whetlier you say
anything — they’ll still bill you. This is the reason why X.25
Chapter Nine Macintosh Media/Cabling
AppleTalk connections have been long in coming: the continual
RTMP traffic made their use uneconomical. X.25 bandwidths range
from modem speeds (1200 baud) to special high-speed connec-
tions at 2 Mbps. Figure 9.26 shows a NetPlCT of a X.25-connected
WAN.
Other Packet Switched Services
In addition to X.25, there are other packet switched technologies
that show promise. Frame Relay is similar to X.25 but will provide
improved performance due to reduced overhead. SMDS (Switched
Multimegabit Data Service) is another packet switched service that
shows great promise. It differs from X.25 and Frame Relay by
emplojang a datagram approach that is more in keeping with the
AppleTalk protocol.
Figure 9.26
11)8 Apple inieinelfouiei
cap be used 10 link two
repioiepeiwoiksoverap
K25pelwofk.
Conclusion
The Macintosh, LaserWriter, and many other peripherals have a
built-in networking capability known as LocalTalk. It is a low-cost
cabling system that works over twisted-pair cabling. AppleTalk
protocols aren’t limited to LocalTalk; they can be sent over
Ethernet, Token Ring, and most other media. While AppleTalk has
been extremely popular on LANs, it has only recently begun to be
accepted as a viable WAN protocol.
“Three”
cintosh is a team player
;Ve^ '.V .X. an' < • . usncf '■ ■ onue
Networks are often used to connect different computer
systems. When it comes to multivendor networks, the
:/!Hcli U‘:>h is a feaw pinye -. I; . u.p
Macintosh is a team player. It supports a wide variety of
rh twui'k Servici , H nnats, Transpc
network Services, Formats, Transports, and Media.
AppJnlafk is also fast 'recoining a d
AppleTalk is also fast becoming a de-facto industry
^ta > - • ' f that is svpj fa ted on ■; )an
Standard that is supported on many different
computers, fron ■ WM PC ro
computers, from the IBM PC to DigitaPs VAX
mincomputer.
frif a wifla variety of
: Services, Formats,
1
<ri.s, and Media.
Living in an
Intei/DOS Worid
acintosh computers and PCs, when connected,
form the most common type of multivendor
network. Exchanging documents and sharing
resources between Macs and PCs is becoming
more commonplace; software developers increas-
ingly are offering separate Macintosh and Win-
dows versions of their applications. With all the
attention and focus on these two platforms, it’s not surprising that
there are so many networking choices available. This chapter will
describe the leading choices for putting your Macintosh computers
and PCs on speaking terms. Since PCs are the most common
network companions, we’ll also use this chapter to introduce some
generic concepts and utilities diat will apply to other networked
peers as well.
Services; Application-Based
During the past several years, Macs and PCs have reached an equal
stature in the eyes of many users. Many applications and services
are supported on both platforms, and since the advent of Microsoft
Part Three Multivendor Networks
Windows 3.0, the distinction between the platforms has narrowed
considerably. Therefore, it’s hardly surprising tliat there’s a strong
desire to share and exchange data between these two machines.
Applications
Many new application programs are being written for both the
Mac and PC platforms. The applications from Microsoft are the
best example. Nearly all their products — ^Word, Excel, Mail,
PowerPoint, Project, and even Flight Simulator — are available for
both platforms. In the majority of cases, the binary file formats of
these applications are the same, and therefore are interchangable.
Other examples of this cross-platform support are Lotus 1-2-3,
Autodesk’s AutoCAD, WordPerfect, and Claris’s FileMaker Pro and
ClarisWorks.
The application support extends beyond simple applications, as
evidenced by Apple’s recent announcement of QuickTime for
Windows and Microsoft’s adoption of TrueType fonts. With the
QuickTime toolkit, Windows developers can create, edit, and
display QuickTime files within their applications. Microsoft’s
inclusion of TrueType might make it a bit easier to move docu-
ments between the platforms without the font alteration that often
occurs when competing font technologies are used.
This cross-platform trend can only increase, as developments are
underway to make it even easier for developers to create either
Windows or Mac versions of their applications through the ad-
vanced object-oriented compilers. These compilers will be able to
substitute the appropriate platform-dependent user interface
elements as necessary.
Other Services
Besides applications, there are other services that can be shared by
both Macs and PCs (see figure 10.1). These include file, print,
Chapter Ten Living in an Intel/ DOS World
database, and mail services. Sharing these services is a somewhat
more complex issue than simply sharing application files. With
many of these services a specific format and transport layer are
often required. For example, a PC can access an AFP file server, but
only if equipped with the AppleTalk protocols. Both Macs and PCs
can access an NFS file server — but only if both machines have the
transport layer of TCP/IP. NFS and most of the other services
common to the UNIX environment can also be installed on a
suitably equipped PC. This chapter will just touch on a few of these
options. For a more complete discusssion of TCP/IP services and
protocols, refer to the next chapter. Regardless of the platform, it’s
often a process of finding a mutually acceptable service that uses
compatible formats, transports, and cabling.
Macintosh PC/Windows
Figure 10.1
Macintosh/PC Services
NelPICI.
Formats: ASCII, EPS, Binary
Compatible
The ultimate success of Macintosh/PC integration is largely
determined by the formats. Fortunately, there are many shared
and compatible file formats. Since the advent of Windows, the
number of applications and formats that are shared by both
platforms has increased dramatically. There are also a number of
mutually acceptable standards, such as Encapsulated PostScript
Part Three Multivendor Networks
(EPS) and Rich Text Format (RTF), that can be seamlessly moved
between the Macintosh and PC platforms.
ASCII
The IBM PC was IBM’s first computer that used the ASCII code
instead of their EBCDIC code. Thus, Macs and PCs speak the same
format when it comes to text files. Both the Mac and PC have
numerous applications that can read and write this format. The
only occasional stumbling block is that sometimes PC text files
have additional line feed characters at the end of each line of text.
These line feeds can often be filtered with special utility programs
or text editors. The key aspect of ASCII is that it’s often used as the
basis for other higher-level formats, such as DXF, IGES, PostScript,
and RenderMan (a 3D photorealistic rendering format). All Mac
and PC text editors and word processors can read and write ASCII
text files.
Word Processing
Many word processors, such as Microsoft Word, usually provide a
built-in translation capability to facilitate the movement of docu-
ments between the PC and Macintosh. Microsoft has also devel-
oped an interchange format called Rich Text Format (RTF) that
attempts to preserve certain formatting attributes between the
Mac and Windows applications. If the PC word processing docu-
ment was not created by an application that also runs on the Mac,
or cannot be directly imported by a Mac program, then it must be
converted. MacLinkPIus, from Data Viz, is a utility program that
Ccm convert most PC word processing formats into formats that are
supported on the Mac. This process is shown in figure 10.2.
Chapter Ten Living in an Intel /DOS World
PC/Windows Macintosh Step 1 Macintosh Step 2 Macintosh Step 3
Of course one option, as shown in figure 10.3, is to save the Mac or PC
document that you wish to transport as an ASCIi fiie. You’li iose ali the
formatting, graphics, and font information, but at ieast you’il have the
document in a form that can be universaliy read. Use this technique as a last
resort.
Hgure 10.2
Converting a PC
Multimaiewoiil
processing document into
alilacintosItMicrosoit
Word document wiilt
MaclinlrPlus.
PC/WIndows Step 1 PCAIVIndows Step 2 Macintosh Step 1 Macintosh Step 2
Figure 10.3
ASCIIislieingusedasa
neutral exchange format
between word processors
on the PC and Macintosh.
Graphics
When it comes to graphics, the situation gets even more compli-
cated. There are dozens of different graphic formats that are
supported on the Mac or PC. Some of these formats are bitmapped
or pixel-based; others are object-oriented, containing object
descriptions (such as line, circle, and rectangle) and coordinate
data.
The leading graphic formats on the Mac are:
• MacPaint-style 72 dpi (dot per inch) bitmaps
Part Three Multivendor Networks
• PICT and PICT2 formats
• EPS (Encapsulated PostScript)
• TIFF (Tagged Image File Format)
Of these, the MacPaint-type bitmap, PICT, and PICT2 formats are
somewhat Macintosh-specific, although many Windows applica-
tions support these file types directly without translation. The
other formats — EPS and TIFF — have become industry standards:
they’re frequendy supported by applications on both platforms.
Just be sure to check the specific applications in question.
On the Windows/ PC side, there are numerous graphics formats,
and with the possible exception of EPSF, TIFF, and some
Windows-specific formats, they are quite often specific to a par-
ticular application. Listed below are but a few examples of the
myriad graphics formats that you’re likely to encounter on a PC,
along with their DOS file extensions.
• Windows bitmap ( . BMP)
• Windows Metafile (.WMF)
• DrawPerfect (.WPG)
• Computer Graphics Metafile ( . CGM)
• Encapsulated PostScript ( . EPS)
• Lotus 1-2-3 Graphics ( . PIC)
• Initial Graphics Exchange Specification (. IGS)
• AutoCAD Drawing Exchange Format ( . DXF)
Chapter Ten Living in an Intel/DOS World
• Micrografx Designer/Draw ( . DRW)
• Tagged Image File (.TIF)
With the exception of Encapsulated PostScript and Tagged Image
File Format (TIFF) files, there are few PC/Windows graphic formats
that are uniformly supported in the Macintosh world. Going from
the Mac to the PC is much easier because support for the PICT
format is common in the PC world; support for the numerous
application-specific PC graphic formats by Macintosh applications
is unlikely.
One solution, shown in figure 10.4, is to use a PC graphics conver-
sion program, such as Hijaak from Inset Systems. It can convert
most PC formats into one of the predominant Macintosh formats
(such as MacPaint or PICT). While Hijaak runs on the PC, the flip
side of that approach is to do the conversion on the Mac by using
the MacLinkPIus graphics translators. They support the conversion
of most popular PC graphics formats into the PICT format.
PC/Windows Step 1 PCAVindows Step 2 Macintosh
Figure 10.4
PC/Winilowsofapbics
lilesaieMsIaiedinio
Maciniostifomaiswiil)
Hiiaakfm Intel
Sysiems.
For CAD/Cy\M and technical graphics, there are two formats that
are widely supported on both the Mac and the PC. The industry
standard IGES is a standard format that can be binary, but is
usually ASCII text. DXF is a standard developed by Autodesk,
Part Three Multivendor Networks
Figure 10.5
PCCADgrapliitsare
iianslaieiiinioMacifliosh
loioiatswilhCADMOVlH
IromKanduSoflware.
developers of the popular AutoCAD program. It too comes in a
binary format, but the most popular format is regular ASCII text.
Both of these formats are commonly supported by both Macintosh
and PC CAD applications and utility programs. These files can
be readily transported between the platforms. In particular,
Macintosh /PC users desiring cross-platform conversion should
examine the CADMOVER translation program from Kandu Soft-
ware of Arlington, Virginia. As shown in figure 10.5, CADMOVER
can read and write ICES and DXF and convert these formats into a
variety of Macintosh graphic formats, such as PICT and EPSF.
PC/Windows Step 1 Macintosh Step 1 Macintosh Step 2
Transports: AppleTalk (PhoneNET
PC). MacIPX, DECnet TCP/IP
Assuming your Macintosh and PCs are physically networked, what
transport protocol will you use to logically connect the machines?
At a certain level the choices are simple. You can equip the PC with
the AppleTalk protocols, equip the Mac with Novell-compatible
protocols, or be a total non-conformist and put foreign protocols
(such as DECnet or TCP/IP) on both machines.
Chapter Ten Living in an Intel/ DOS World
The Choices
The choice of a suitable transport protocol is often complicated by
the fact that not all services support ever)' transport. Therefore, tlie
selection of a transport often begins at the top of the stack, at the
Service layer. If the desired service is AFP file services, then you’re
somewhat limited to choosing AppleTalk as a transport. In most
cases, there aren’t too many choices when it comes to transports.
All too often, a given service only supports a single transport
protocol. The basic choices are simple: Either make the PC speak
the Mac transport of AppleTalk, make the Mac speak a PC trans-
port such as Novell IPX, or make both the Mac and PC speak a
“foreign” protocol such as TCP/IP or DECnet. The only other
options are to use a gateway or use an intermediate computer to
act as an exchange agent. Let’s explore the options.
AppleTalk on the PC
If you have a network where Macs are predominant and there are
just a few PCs, then consider turning those PCs into AppleTalk
nodes. One way to do this is with Farallon’s PhoneNET PC. It
equips a DOS or Windows PC with the AppleTalk protocol stack
and also provides the equivalent of the Chooser (see figure 10.6), so
that the user is able to select AppleTalk network services (such as
AFP file services. System 7 File Sharing, and PAP print services).
PhoneNET PC works either with a LocalTalk/PhoneNET card
(offered by Apple, Dayna, COPS Inc., and Farallon), most common
PC Ethernet cards, and IBM Token Ring 16/4 cards. Figure 10.7
shows various NetPICT diagrams of a PC running PhoneNET PC
over different cabling mediums.
Part Three Multivendor Networks
Figure 10.6
Farallon'sPiionefiEIPC
pals die AppleTalk
proiocolanilaCliOQser’
eqaivalenioatotliePC.
Figure 10.7
farallon'sPhoaellETPi:
suppans AppleTalk aver
dilleieni kinds al cabling.
3
Chooser C
Nttwoffc servlets: Select a server:
AppleTalk Zones:
4th Floor
Sth Floor
6th Floor
gavel and
AppleShare
ARA DIalln
ElsO
Open Window
PaparLata
R&D
Coimact
PLESHAREMPE0PLEJ^4
MKTG LASERWRITER NTX (lASE
MKTG LASERWRITER NTX (EPSON)
n
PC/Windows
Macintosh
AFP & PAP
Services
AFP & PAP
Services
Formats
\ / Formats
AppleTalk
I 1 AppleTalk
> Cabling
\ Cabling <
In addition to the AFP and PAP services supported by PhoneNET
PC, various application services are supported by the product as
well. A listing of the currently supported PC application appears
below.
• Farallon Timbuktu (PhoneNET PC is included with all ver-
sions of Timbuktu for Windows)
Chapter Ten Living in an Intel/DOS World
• Microsoft Mail 3.0 and 3.1
• QuickMail from CE Software
• Lotus cc:Mail
• WordPerfect Office
• Claris FileMaker Pro 2.0
• Blyth Omnis 7 (Can access DAL services)
• Microsoft FoxBase+
As shown in figure 10.8, putting PhoneNET PC on a PC doesn’t
restrict it from running other protocols, as PhoneNET PC runs
concurrently with NoveU NetWare IPX and TCP/IP. One advantage
of turning your PCs into AppleTalk nodes is that, like the
Macintosh, your PC wall self-configure its network address. Your
PC will also be visible to AppleTalk network management and
troubleshooting utilities such as Apple’s Inter»Poll.
Rgure 10.8
Faialion'sPhonellEIPC
sappoiisiheconcuiieni
PSP ol Novel IPX and
ICP/IPinaddilionio
ApplelallcitissuppDiied
onPClocallalkcatds,
Elheinel and Men Ring.
Part Three Multivendor Netwrorks
In addition to the Farallon products, another company,
Cooperative Printing Solutions Inc. (COPS Inc.), has a number of
AppleTalk PC solutions. COPSTalk for DOS and COPSTalk for
Windows puts the AppleTalk protocols on a DOS- or Windows-
equipped PC. Other products include PServe, which provides
unified print queuing for Macs and PCs, and the COPS EasyServer,
which turns a PC into an AFP file server.
Yet another PC AFP solution is MacLAN Connect Gold from
Miramar Systems Inc. This product connects LocalTalk, Ethernet,
or Token Ring users to file and print services. It also supports the
recently announced Microsoft Windows for Workgroups.
Novell Solutions
Novell, the networking giant, offers two approaches to Macintosh
PC connectivity. First, Novell NetWare servers can be configured as
AFP file servers. As shown in figure 10.9, Macs access these services
with the AppleTalk transport; PC users access the same server with
Novell IPX protocols. Normally it’s not recommended, but it is
possible for the two environments to share a common file space.
Thus, it is possible to share files between the two environments.
Rgure 10.9
Novell's NetWare lor
Maciolosltadds
AoplelalUFP.andPAP
suooorllo PC/Wiodows
cooipoters.
Chapter Ten Living in an Intel/DOS World
Dayna offers an alternative for the Novell environment, one that
turns a Mac into a full-fledged NetWare client running the Novell
IPX protocol. As shown in figure 10.10, with Dayna’s NetMounter,
the AppleTalk protocol stack is replaced with a version of Novell
IPX. A Chooser-level driver is used to select and log on to the
server. With this approach, the Mac appears as a PC client, so you
won’t have to purchase the NetWare for Macintosh module.
NetMounter also includes a utility that will assign Macintosh
creator and type codes to particular DOS file extensions. This
makes it possible to copy a Microsoft Excel spreadsheet from a
NetWare DOS server and have its Macintosh icon available for
subsequent double-clicking.
PC/WIndows PCAWIndows Macintosh
Figure 10.10
Dayna'sNeiMouniet
equips Macs Willi (lie IPX
tianspoitlayeianil
enables ihem 10 direclly
access any NeiWaie
server.
in the past, Novell concentrated on the file service support for the
Mac, but Novell has recently begun to develop another approach.
MaclPX, shown in figure 10.1 1, involves placing Novell’s IPX
protocol on the Macintosh. Intended for developers, MaclPX will
provide an avenue for cross-platform peer-to-peer communica-
tions between applications on the two platforms. MaclPX should
help the Macintosh make inroads into environments where PCs
and Novell networks prevail.
Part Three Multivendor Networks
Figure 10.11
Novell's MacIPXKanspoit
piovldescioss-plaiforoi
peei'iO'peer access.
Another Novell Macintosh solution is DataClub, which was re-
cently acquired by Novell from International Business Software
Inc. DataClub is a distributed file service that transparently aggre-
gates and manages disk space from the individual workstations.
This virtual disk space is made available to the network users as
mountable disk volumes. DataClub was originally developed for
Macintosh networks, but Novell plans to add support for DOS,
Windows, and UNIX.
Banyan Vines
Banyan Vines, a popular enterprise-wide networking system, offers
client support for Macintosh workstations, allowing them to share
resources with DOS, OS/2, and Windows workstations on the same
network. Vines supports the AppleTalk filing protocol and the
Printer Access Protocol from their Intel /PC-based Vines server. In
addition, AppleTalk tunneling through Vines enables disjointed
AppleTalk networks to be connected with each other via the Vines
network.
DECnet on the Mac and PC
While the appeal of DECnet is somewhat limited to those sites with
DEC VAX computers, another transport alternative is to use
Chapter Ten Living in an Intel/DOS World
DECnet. Wlien DEC got into the PC clone business, they quickly
saw the potential of integrating PCs into their network environ-
ment. To achieve this goal, DEC created DECnet DOS, which
equips the PC with the DECnet protocol stack. They also developed
a number of Service layer applications so that the PC user can
engage in terminal emulation, file copying, and printing within a
VAX computing environment.
DECnet for the Macintosh is now offered as a component of
PATHWORKS for Macintosh, the product that grew out of the
Apple/DEC alliance. DECnet for the Mac performs a similar
function to its DOS counterpart and enables a Mac to perform
services such as file copjang. E-mail and terminal services in a
DECnet environment.
Because both the Mac and PC support the DECnet transport
protocol and the file copy services (see figure 10.12), they can use
this mechanism to move files between the two platforms, even
when there’s no VAX involved. It should be noted, though, that the
common services offered by tlie Mac and PC versions of DECnet
are somewhat basic and are likely to be of interest only to those
who already have VAX computers and DECnet neUvorks already in
place.
DEC VAX PC/Windows Macintosh
Figure 10.12
BoiiuheMacandPtcaii
use Digital’s DECnel
pioiocol, either will) a
VAX or separately.
Part Three Multivendor Networks
Figure 10.13
HielCP/IPiransponis
suppoiiedoninost
plailorms today, ipcluding
the Mac and PC. Sun's
Netwfltkfilinp System is
a common tile service
supported by TCP/IP.
TCP/IP on the Mac and PC
If you're still looking to put a “foreign” transport protocol on your
Macs and PCs, then another choice is TCP/IP. Common in UNIX
circles, TCP/IP stands for Transmission Control Protocol/Internet
Protocol. It is widely used in universities, government installations,
defense contractors, and many large corporate installations.
MacTCP is an Apple product that equips the Macintosh with the
TCP/IP protocol. MacTCP is similar to AppleTalk, in that it’s
useless without application services. Most of the application
services that work with MacTCP are provided by third parties. The
popular services are NFS file service, SLIP, and TELNET terminal
services, and X-Windows. These services sit atop the MacTCP
transport layer.
Using file access as an example in figure 10.13, a Macintosh
equipped with MacTCP and NFS (available from either
Wollongong and InterCon) can exchange files with a similarly-
equipped PC. PC-NFS from Sun Microsystems is a popular imple-
mentation of NFS for DOS. FTP Software’s product called PC/TCP
Plus provides the PC with the TCP/IP transport protocol and NFS
in one package. For more information on TCP/IP solutions, refer to
the next chapter, which covers UNIX connectivity.
Chapter Ten Living in an Intel/DOS World
Cabling Options
The choices available for cabling systems for Mac/ PC connectivity
are as varied as the cabling choices available for the Macintosh.
You can choose among LocalTalk, Ethernet, Token Ring, and even
the popular PC cabling .system, ARCNET.
LocalTalk
The benefits of LocalTalk expressed earlier apply to PCs as they do
for Macs. It’s cheap and it runs over conventional tvWsted-pair
phone wire. If you’re introducing PCs into a largely LocalTalk
network, you’ll be able to equip your PCs with the LocalTalk
hardware. Vendors such as Dayna, Farallon, and COPS Inc. offer
such cards. Versions for either the ISA or MicroChannel bus are
available.
It probably only makes sense to consider LocalTalk cabling for
AppleTalk-only networks. If you’re planning a multiprotocol
network, you’re better off with Ethernet or Token Ring cabling. You
will also find that the LocalTalk cards for PCs (see figure 10.14) are
nearly the same price as the PC Ethernet cards.
Macintosh
V
LocalTalk
I
PC/Windows
LocalTalk
"j
)
Figure 10.14
locallalkonaPCKetPICl.
Part Three Multivendor Networks
Figure 10.15
EtoneionaPCNeiPICI.
Ethernet
Ethernet is probably the best choice for a PC/ Mac cabling system
(see Figure 10.15). It’s relatively inexpensive, comes in a variety of
physical cable types, and (most importantly) it supports a wide
variety of networking transport protocols. If you’re planning to use
multiple protocols, just be sure that all your anticipated protocol
drivers support your chosen Ethernet cards. As mentioned before,
PhoneNET PC from Farallon supports most Ethernet cards, such as
the EtherLink series from 3Com and Novell’s NE1000/NE2000, but
you may Find a card that is not supported. Just be safe; double-
check with all vendors and make sure everything’s supported.
Macintosh PC/Windows
Token Ring
Compared to Ethernet, there’s less variety in Token Ring cards.
IBM is becoming very agressive in pricing and licensing their
Token Ring componentry (see Figure 10.16). Prices have already
started to drop. Most of the newer cards offer both 4MB and 16MB
transmission rates and use tlie latest integrated circuitry. As with
Ethernet, just make sure that the transports you’ve selected are
supported. Token Ring-equipped PCs are likely to be popular in
IBM mainframe and minicomputer environments where Token
Ring is already in use.
Chapter Ten Living in an Intel/DOS World
Macintosh PC/Windows
Figure 10.16
Men fling on a PC
NeiPICI.
Apple offers a NuBus Token Ring card. This can be used to connect
a Macintosh to a Token Ring network directly, or it can be used by
a Mac running the Apple Internet Router. When a Mac runs the
Apple Internet router, it can be equipped with other networking
connections such as LocalTalk and Ethernet. In this case, the
router can be used to protide LocalTalk- and Ethernet-equipped
Macs access to services on the Token Ring network.
ARCNET
Shortly after the IBM PC was introduced, the ARCNET network
system was designed as a low-cost LAN (see figure 10.17). ARCNET
is still in widespread use today as a cabling medium for PC net-
works, particularly NetWare. ARCNET supports two different kinds
of cabling (coaxial and twisted-pair) and has a bandwidth of 2.5
Mbps. This places ARCNET betw'een the bandwidths of LocalTalk
and Ethernet.
In 1991, ACTINET Systems began to offer a line of ARCNET cards
for Macintosh computers. Their ARCTalk cards come in three
models: one for NuBus-equipped Macs, one for the Macintosh LC,
and one for the SE/30 and Ilsi. Included with the cards is ARCTalk
software which provides AppleTalk Phase 2 support. Although the
original ARCNET specs permit both star and bus topologies, the
ARCTalk cards only support the star topology.
Part Three Multivendor Networks
Figure 10.17
AflCNEIonaMacaniiPC
NeiPICI.
Macintosh PC/Windows
With the coaxial version of ARCNET, each cable run can be a
maximum of 2000 feet. The twisted-pair implementation can run
400 feet (slightly longer than twisted-pair Ethernet). In bodi cases,
an ARCNET hub, available from third-party vendors, is required.
The cards are certified by Novell for use with NetWare for
Macintosh and the cards and ARCTalk software are System 7-
complient and wili run on the Apple Internet Router. Use of the
Apple router is important because there are no options for con-
necting LaserWriters or other LocalTalk devices to an ARCNET
network.
Conclusion
There are more networking services, formats, transports, and
cabling options available to connect Macs and PCs than with any
other platform combination. The key is to first determine the
desired services. Then, examine which protocol best delivers that
service. Finally, choose a cabling system that balances support for
the chosen protocol or protocols with cost and future growth and
expansion.
UNIX
Connectivity
11
® hile not as prevalent as the PC or the Mac, UNIX-
based workstations and other UNIX computers
are very common, particularly in the engineering
and technical environments where Macintosh is
also very' popular. This chapter explores the
networking options for Macintosh and UNIX
Services: FR NFS. AFR
X-Windows, TELNF
Most of the services in the Macintosh/UNIX world revolve around
file access and terminal services. To a seasoned Macintosh user,
the services offered in the UNIX-TCP/IP world may seem a bit
basic. Many of these services are directed at sophisticated users
and programmers. However, new TCP/IP-conversant Macintosh
services aimed at a more general audience are being announced
every week.
Part Three Multivendor Networks
Rather than starting off talking about bringing UNIX services to the
Mac, we’ll begin by looking at the AppleTalk-based services
available for the UNIX environment.
Figure 11.1
Xinel'sI-AShareiuins
a Sun or HP UNIX
workstation into
an AfP Server.
AFP/PAP Services on UNIX Computers
To a Macintosh user, file service means AFP. There are several
products that provide this service on UNIX computers. Xinet sells
K-AShare, which implements 7VFP on a Sun or HP UNIX host (see
figure 11.1). So, just like with any other AFP server, the Macintosh
user accesses UNIX files through the Chooser. The mounted
volumes provide access to the UNIX files as if they were local disk
files on the Mac. (The Sun or HP running K-AShare could also be
set up as an NFS server, so in a sense K-AShare can be thought of
as an AFP/NFS gateway.)
1 ^
Macintosh Client Sun Workstation
A companion product from Xinet, K-Spool, provides PAP print
spooling services for the Macintosh user. It also acts as a print
server and PostScript agent for the UNIX user, forwarding print
jobs to a networked LaserWriter.
Diagrammed in figure 11.2, K-Spool’s PAP spooler appears to the
Macintosh user just as any other LaserWriter. It does this with a
Chapter Eleven UNIX Connecti\nt>'
special program — a virtual LaserWriter. This means that the
program accepts and responds to NBP Lookup Requests and
Responses. After the NBP process is done and a Macintosh user
selects the virtual LaserWriter, the program accepts the PAP
commands that contain the PostScript printing instructions. This
printing information is placed in a file that is queued until the
designated printer is available. This technique avoids the annoying
background delays seen when Print Monitor is used to spool the
print file on your local Macintosh disk.
@ 5 ?
This program, or This program, or
virtual LaserWriter, virtual Macintosh,
appears to the Mac appears to the real
user as a real LaserWriter as a
LaserWriter. real, printing, Mac.
Rgure 11.2
Xinei's ( Spool turns a
Sun or HP UNIX
workstation Into a PAP
print spooler.
Pacer Software offers PacerShare, shown in figure 1 1.3, which
offers a similar capability for VAX computers equipped with Sun,
HP, and DEC Ultrix (DEC’S version of UNIX). PacerShare, like K-
AShare, maps the AFP file structure onto the native UNIX file
structure of tlie host. PacerPrint provides for LaserWriter printer
sharing.
InterCon’s InterPrint is another printer utility that provides
Chooser-level access to UNIX printers on TCP/IP networks. This
utility makes it as easy for a Mac to print to a UNIX PostScript
printer as it is to print to a LaserWriter.
Part Three Multivendor Networks
Figure 1 1 .3
PacetelorimiX
piovides AFP file seivices
oiiSun.llP.aiiilDECIIIim
VAXcompulers.
Macintosh Client Sun, HP, VAX Ultrix
Figure 11 .4
Apple and Pacef provide
DAI dalabase seivices on
UNIX Gompuiers. All
UNIX DAI seiveis use
AppleTalk as a liansppft
Data Access Language
DAL services provide access to relational databases on UNIX
systems. Apple sells a version of DAL for A/UX equipped Macs, and
Pacer offers a DAL server for HP UNIX computers, DEC Ultrix, and
Sun’s SPARCstation (see figure 1 1.4). These services use the
AppieTalk transport on each of their respective platforms.
Macintosh or PC Client UNIX Host
Chapter Eleven UNIX Connectivity
File Transfer Protocol (FTP)
The File Transfer Protocol (FTP) is commonly used in UNIX and
TCP/IP networks to move files between nodes. FTP provides
directory services, so a user can get a listing of candidate files on a
remote machine. It also provides support for a variety of formats
(such as ASCII and binary) and insures security by requiring login
IDs and passwords. However, some systems offer public access to
files witliout IDs and passwords. This service is known as Anony-
mous FTP, and many companies, universities, and organizations
use it to post information that they wish to distribute freely.
Most implementations of FTP are through terminal-based com-
mand line interfaces; any of the TCP/IP compatible terminal
services (such as VersaTerm-Pro and NCSA Telnet) can be used to
engage in FTP transfers (see figure 1 1.5). Since the Macintosh has
focused attention on the user interface, a number of FTP applica-
tion services are starting to provide sophisticated interfaces.
There’s even a freeware HyperCard stack called HyperFTP, written
by a Cornell student, that supports the basic set of FTP operations.
Macintosh UNIX Host
Figure 11.5
Iermin3lemulaiofs.such
asUeisaleimPro.
piovideFIP services to
MaclCPequIppeilMacs.
Part Three Multivendor Networks
Hgure 11.6
FIPSliaieliDmAilvanGeil
SollwareConceplsuses
FIP 10 provide Macs ivi
aFileSFiariog-lilre
capaliiliiv.
FTPShare, from Advanced Softvc'are Concepts, is a multi-session
FTP server (up to 20 simultaneous sessions) that runs in the
background of the Macintosh (see figure 1 1.6). In many ways,
FTPShare can be compared to Apple's System 7 File Sharing. The
FTPShare setup screens, shown in figure 11.7, resemble the screens
used under System 7 File Sharing, so it should be easy for
Macintosh users to adapt. FTPShare also includes a monitoring
application drat monitors the connected users and gives an indica-
tion of FTP activity.
Figure 11.7
FlPSliaie puts the look
aud Feel oFMaciniosli File
Sharing onto FIP
iranslers.
0 Jeff 192.9.200.13 Dlog
18? Connected Users Q Folder
enoegteem
1929200.1
pi*Ne
2
"K
FTPS
I FTP Share Setup !
ilE-iimga.at iluHu; r-rr ^
ggSgSH — I [ Disconnect ] [ Trace )
- Netwurk Identity
0«m«r
O^mr Pusword : |— —
IP M4r*iS
192.9.200.13
P
FTP Shore
[ ] [g
[ Settings
FTP Shtr* (< on. Clkk Stop to prtvtnt ottt«<
U10T5 from Mowing ttitlr Cntrg FoMor.
enonymout I
wl.O fron Jeff on Moo, Jon 22, IW2, 4.53:4
1
In. you don‘1 hooe Urite Prioileyu
W, I. IS, 17
d ok.
ted succesful ly.
nU,U,'Au, I, IS, 18
Chapter Eleven UNIX Connectivity
Network Filing System (NFS)
Developed by Sun Microsystems, NFS provides additional capa-
bilities not found in FTP. It can be used to transfer Files between
computers, but it also provides a mechanism for distributed
applications that are network-aware. There are several implemen-
tations of NFS on the Macintosh; two of the more popular products
are NFS/Share from InterCon and Pathway NFS from Wollongong.
Both of these products, shown in figure 1 1 .8, work with Apple’s
MacTCP and bring NFS services to the desktop of the Macintosh,
thus retaining the ease-of-use of the Mac.
Figure 11.8
IniefConandWollongonD
ollefNlSseiviceslorilie
Macteli. Ihey boll) use
iheMacICPiianspon
protocol.
Partner, from IPT, takes another approach to Macintosh/UNK
connectivity. It works in both directions between a Macintosh and
Sun SPARC systems. Macintosh users access SPARC files as AFP
server volumes; SPARC users access the Macintosh files as if they
were NFS volumes (see figure 11.9). IPT also provides PAP print
spooling with a variety of UNIX print services and a mail gateway.
The Partner application resides on the Sun SPARC machine and
does all the necessary transport and service layer conversion; no
special software is required on the Macintosh.
Part Three Multivendor Networks
Figure 11.9
IPI's Fanner piovidesbi-
tlireclional access of Mac
and Sun SPARC tiles over
lire network. Mac users
seeilieSPARCsiationas
an AFPserver, while Sun
users see the Mac as just
another NFS server.
r
Macintosh
3:
Sun SPARCstation
Although NFS was developed by Sun, it is supported on many
different computing platforms. Of course, it’s most common on
UNIX computers, but can also be found on PCs (PC-NFS), DEC
VAX and IBM mainframes and minicomputers. Any NFS-equipped
Mac should be able to access these other NFS platforms as well.
X-Window
The X- Window standard, also called X or XI 1, evolved from initial
work done at the Massachusetts Institute of Technology. It is a
client/server windowing environment that is extensible and
customizable. DEC’S implementation of X, known as DECwindows,
is used as their standard windowing environment on their VAX/
VMS computers and UNIX-based workstations. Motif is the name
of the X implementation by the Open Software Foundation (OSF).
However, the concept and underlying messaging used by these
different flavors of X is the same. The user interfaces are slightly
different, with different styles of windows, buttons and controls.
Chapter Eleven UNIX Connectivity
X is popular in UNIX (and tlierefore TCP/IP) networks. However,
it should be mentioned that X is not limited to computers run-
ning the UNIX operadng system or the TCP/IP transport. The X-
Window service can be run on many different kinds of computers
and network transports.
X works on a client/server model. The computer that displays the
image is called the server and the computer that runs the applica-
tion code is called the client. This is a little confusing because we're
accustomed to calling the desktop device the client, but with X
the terms client and server reflect how the underlying software
operates.
Many companies make X- Window terminals, or servers. These
stand-alone devices have just enough processing power to run the
X server display software (see figure 11.10). These terminals
connect to X client programs running on other computers. Most
UNIX workstations can act as both X clients and servers. Some-
times the client and server even run on the same machine, but the
system is flexible enough to permit connections locally or re-
motely.
Figure 11.10
XWiniiowteiiiiinals,or
display senms.«ioik
iptofljunciionwiih
X'Wjpd(iwclienis.Ilie
diems aieHieapplicaiion
programs, viliicli use llie
display servers to do
ilreirgrapliicalwork.
X'Window diems and
servers can be
disiribuiedoveta
network, or logeiber 00
the same computer.
Part Three Multivendor Networks
There are two X- Window servers for the Macintosh: MacX from
Apple and eXodus from White Pine Software. Either one can be
used to connect to X- Window client applications that run on UNIX
computers (or any other platforms that support the X standard).
These applications make it possible to display X- Window applica-
tions side-by-side with Macintosh windows.
Planet X, from InterCon, is an X- Window client for the Macintosh.
It allows a user to view and operate a Macintosh remotely from
any X- Window server. As shown in figure 11.11, this server could
be a UNIX workstation or even another Mac running MacX. When
a session is established, the user operating the X- Window server
sees the Macintosh desktop in a separate window. Planet X sup-
ports the cutting, copying, and pasting of graphics between the two
environments.
Figure 11.11
Plane! X is die opposite of
MacX.l!!urnsaUacinio
an X-Window client,
ntakinp it accessible liom
oiheiX'Windowseiveis.
Macintosh
UNIX
For all its flexibility, X has yet to really take off. The number of X
applications continues to grow, but at a slow pace. The bulk of X
applications can be found in the scientific and technical world
where UNIX is an established standard. X is also reasonably
Chapter Eleven UNIX Connectivity
popular in its DECwindows guise, where Digital continues to
migrate its traditionally VT terminal-based applications over to X.
Terminal Services
The most basic of TCP/IP services is the Telnet terminal service.
This is a simple protocol that allows a user to remotely connect to a
server as if he were using a local terminal. There are several popu-
lar terminal emulators that offer Telnet and work in concert with
MacTCP (see figure 11.12). Two examples are NCSA Telnet and
VersaTerm Pro. With these applications, you can log on to remote
TCP/IP hosts over a dial-up connection, or over a network connec-
tion such as Ethernet.
Macintosh UNIX Host
Figure 11.12
lelnei is a popular UNIX
lerminal service. Many
Macinlosli terminal
emulaiors support it
using MacICP as the
transport
Other terminal services use the Macintosh Communications
Toolbox (MCT) with MacTCP to provide other kinds of terminal
services. Advanced Software Concepts has an entire family of
terminal emulation products that work through MacTCP. asc5250
is a MCT terminal emulation tool that allows you to communicate
with IBM AS/400 minicomputers over a TCP/IP network. asc3270
is used to enable Macs to communicate with IBM mainfi'ames,
using MacTCP (see figure 1 1.13). This seems weird — ^TCP/IP has
been associated with UNIX workstations for quite a while — but it’s
Part Three Multivendor Networks
becoming more commonplace as TCP/IP begins to displace IBM's
non-routable SNA protocol.
Figure 11.13
NelPICI of an unlikely
irloiMaciniosliis
funning ASCs IBM 3270
lermioalemulaioi using
MacICPasaifanspnft
Macintosh IBM Host
Formats
similar to the formats supported by the PC, UNIX computers are
able to share ASCII text files and those application formats that are
binary compatible. As an example, let’s imagine we have three
computers — a Mac, a Sun workstation, and a PC networked with
Ethernet. Autodesk makes a version of their AutoCAD CAD pro-
gram for each of these computing platforms. Each computer has
TCP/IP and NFS file ser\dces installed. Since AutoCAD uses identi-
cal binary formats for all platforms, we’re able to freely move
AutoCAD files among the different computers. No translation will
be required.
Taking this scenario further, imagine we were using Claris CAD on
the Macintosh instead of AutoCAD for the Macintosh. Unfortu-
nately, Claris CAD can’t directly read the binary format of an
Chapter Eleven UNIX Connectivity
AutoCAD file. In this case we’ll have to translate. We can generate
an ASCII-based DXF file from the AutoCAD application on the Sun
or the PC (see figure 11.14). We then use NFS file services to
transfer the DXF file to the Macintosh, where we use the Claris
Graphics Translator to convert the DXF format into the native
Claris CAD document.
Sun Workstation
AutoCAD
DXF
Transport
Cabling
Macintosh Step 1
Macintosh Step 2
Ctaris Graphics
Claris Graphics
Translator
Translator
DXF
\ / Claris CAD Format
Transport
I 1 Transport
V Cabling
/ S Cabling <
E
Macintosh Step 3
Claris CAD
Application
' Claris CAD Format
Transport
Cabling
Many UNIX workstations, particularly those firom NeXT, use
PostScript as an imaging format. PostScript files from these ma-
chines (including the EPSF variety) can be easily exchanged with
the Macintosh (see figure 11.15). But, as mentioned before, be
aware that EPSF files generated by non-Macintosh applications
may not have the PICT preview resource.
NeXT Workstation
Macintosh Step 1
Macintosh Step 2
Macintosh Step 3
Figure 11.14
A DXF file generated Oil
a Sun workstation is
converted into a Claris
CAD document Willi tlie
Claris Giaplticslranslaior.
Figure 11.15
A PostScript lile (without
the Pin preview! is
generated by a NeXI
application, moved over
lolheMacinlosli
translated into a
PostScript hie with a PICT
preview, and then placed
into a PageMaker
document.
Part Three Multivendor Networks
Transports: AppleTalk, TCP/IP
Mention UNIX networking to most computerphiles and a discus-
sion of TCP/IP will undoubtedly ensue. TCP/IP is an important
part of Macintosh/UNIX networking. MacTCP, from Apple, puts
the TCP/IP protocol stack onto the Macintosh, giving the
Macintosh peer-to-peer capabilities with other IP nodes. The other
option is to equip the UNIX machine with AppleTalk protocols.
Computers so equipped can serve as AFP file servers, PAP print
spoolers and DAL database servers.
AppleTalk for UNIX
Versions of AppleTalk for UNIX have been implemented by Xinet
and Pacer (see figure 11.16). These products dovetail with each
vendor’s respective services (listed previously). Xinet’s AppleTalk
protocol suite is called K-Talk and is the underlying protocol for
their K-AShare AFP server and their K-Spool PAP spooler. Pacer’s
AppleTalk implementation is used for their PAP and DAL imple-
mentations. Both implementations support AppleTalk Phase 2 and
must be configured in accordance with the AppleTalk Phase 2
network numbering rules and zone naming conventions.
Hgure 11.16
Ilie AFP file slianniL PAP
piiQi spooling, and DAI
daiabasesaiYicesoHered
by Xinet and Pacer are all
delivered via Applelalk
proiocols.
TCP/IP on the Macintosh
The flip side of AppleTalk on UNIX is TCP/IP on the Macintosh.
Apple’s MacTCP is widely supported by many third-party vendors
Chapter Eleven UNIX Connectivity
listed in this chapter — it’s the engine that is used by these vendors
to create their applications. Figure 11.17 is a diagram of MacTCP
transport protocol.
Terminal, File,
Print, Mail Svcs
Formats
MacTCP
TCP/IP
Cabling
Figure 11.17
MacKP is Apple's
implementaiioppfICP/IP
prpipcolsfonhe
Macipiosh.
If you’re familiar with TCP/IP networks, configuring MacTCP is a
snap. If you’re a Mac person and a TCP novice, be prepared for a
different world. TCP/IP nodes can obtain their addresses manually
or by a dedicated server. There is a mechanism for dynamically
assigning addresses, but it’s somewhat limited (unlike AppleTalk).
TCP/IP addressing, also referred to as IP addressing, uses 32-bit
addresses. Unlike AppleTalk addresses, IP addresses can alter the
numbers assigned to the network and host identifiers. There are
five classes of IP addresses, Class A through E, of which classes A
through C are commonly used.
The fundamental difference between the classes is whether you
create an address range that provides a small number of networks,
each with many nodes; or a range that provides many networks,
each with a limited number of nodes. Class A addresses use 7-bit
network numbers and 24-bit host numbers. This adds up to 31 bits.
(The first bit of a Class A address (0) is reserved and is not used as
part of the address.) This provides 128 possible networks, each
containing more than 16 million hosts. Class A networks are rare
and are used for very large IP networks.
Part Three Multivendor Networks
Class B addresses split the 32 bits differently, by using 14-bit
network numbers and 16-bit host numbers. (Here, the first two bits
(10) are reserved.) Class B networks are common, since they strike
a balance between the number of networks and the number of
nodes.
Class C addresses use 21-bit network numbers and 8-bit host
numbers. The first three bits (110) are reserved. Class C addresses
are similar to AppleTalk networks, in the sense that both number-
ing schemes are based on a large number of networks with a
limited number of nodes per network.
The 32-bit IP address is usually written in a four-part decimal
notation known as a dotted quad; an example is 1 31 . 43 . 3 . 1 8.
Each part is eight bits (also called one byte or octet) long. With a
Class A address, the first octet (minus the one reserved bit) repre-
sents the network ID; the remaining three octets are the host ID.
Class B addresses use the first two octets for the network ID (minus
the first tv\'o bits). The remaining two octets are used for the host
ID.
Class C addresses use the first three octets for the network number
(less the first three bits). The remaining octet is used for the host
ID.
.<\n exact analogy to AppleTalk addressing is difficult because IP
addresses do not specify a particular computer. Since IP addresses
contain network numbers and host numbers, they cannot repre-
sent single computers that may have multiple connections to
different networks. Therefore, an IP router or gateway will have an
IP address for each network connection — not each computer.
The name TCP/IP comes from the functions of different parts of
the protocol stack. The Transmission Control Profoco/ performs the
same tasks as the AppleTalk protocols at the Session and Transport
Chapter Eleven UNIX Connectivity
layers of the OSl Reference Model (see figure 11 . 18 ). TCP ensures
the reliable, sequenced delivery of datagrams. The IP part of TCP
stands for the Internet Protocol. This corresponds to the Network
layer of the OSl model — the layer where AppleTalk DDP protocol
resides. The function of this layer is the same in both cases; both IP
and DDP are responsible for addressing the source and destination
of the message.
NetPICT OSl Reference Model TCP/IP
Services
Application
Application
Format >
Presentation
PostScript, FTP,
NFS.TsInst
Protocol
Session
Transmission Control Protocol
Transport
Transmission Control Protocol
Network
Intarnst Protocol
^ Cabling
Data Link
Data Link
Physical
Physical
Figure 11.18
ICP/IP gels ils name limn
iwopioiocolsUiai
tmiesgoniliosgecilic
aieasolilieOSI
Defeience Model.
AppleTalk DDP datagrams are addressed with network, node, and
socket numbers of the source and destination: for example:
DDP Datagram
From: 12.22.128
To: 14.43.129
IP Datagrams are addressed with the network and host numbers of
the source and destination;
IP Datagram
From 128.10.2.24
To: 191.5.48.12
Part Three Multivendor Networks
Cabling: Ethernet, LocalTalk,
Token Ring
By far, the majority of the UNIX workstations and computers rely
on Ethernet as a cabling medium. There are the exceptions, such
as an IBM AS/400 speaking TCP/IP over a Token Ring network. For
the most part, though, Macs connecting to UNIX computers will do
so over Ethernet.
MOTE: It’s important to remember that the installation of MacTCP won’t
preclude the use of AppleTalk protocols over the same network connection.
With one Ethernet card, you’ll be able to run the AppleTalk and TCP/IP
protocols concurrently.
Figure 11.19
locallalkUacs
coniinunicaiinowiili
Anplelalk-eQuippeilUNIX
compuierspnltheinel
canuseapylocallalk-lo-
Iiliemeiiouiei.
If you decide to choose an AppleTalk-based service, you have the
choice of LocalTalk, Ethernet, or Token Ring as the cabling me-
dium. If you choose LocalTalk, remember that you’ll need a router
that passes AppleTalk from LocalTalk to Ethernet (see figure 11.19).
If you choose TCP/IP on the Mac, you’ll be able to use LocalTalk,
Ethernet, or Token Ring, but if the UNIX computer doesn’t have
the same cabling interface you’ll need the appropriate IP or
multiprotocol router to make the connection (see figure 1 1.20).
Chapter Eleven UNIX Connectivity
Figure 1 1 .20
locallalk Macs using
TCP/IP 10 communicaie
will) UNIX coinnuieis on
Eltalmosiusea
locallalk-iO'Eitiernei
[ouleriliaisupponsIP
looting.
Conclusion
The Macintosh can connect to UNIX computers at all levels. UNIX
computers can speak the AppleTalk protocol, Macs can speak the
TCP/IP protocol, and most UNIX computers and workstations
utilize Ethernet cabling, so the integration process is often as
simple as installing the right software.
DEC VAX
Connectivity
efore the famous Apple- IBM Alliance, there was
tlie Apple-Digital Alliance. The collaboration
between these two companies resulted in an
architecture that combines the best of both
worlds. Even though both the VAX and the
F Macintosh have often been criticized for their so-
called “closed architecture," the PATHWORKS for
Macintosh environment illustrates that an open architecture is
more than a common networking protocol or operating system.
Senrices: PATHWORKS
fat Macintosh
The services offered by DEC’s PATHWORKS for Macintosh are a
combination of products developed by Apple, Digital, and third-
party vendors. They address the basic requirements of terminal
emulation, X-Window (DECwindows) emulation, AFP File services,
PAP print services, DEC print services, DAL database services, and
E-mail.
Part Three Multivendor Networks
Figure 12.1
Macleiminal provides
l/II02aedVI320ieroiiaal
efaplaiionlorUaciniosli
users connected 10 DEC
VAXconipulers.
MacTerminai
Traditionally, VAX access has been accomplished with terminals.
The ubiquitous VTIOO terminal standard (and subsequent versions
that are found in nearly all Macintosh terminal emulators) is based
on DEC products. MacTerminai is an Apple terminal emulator that
comes bundled with PATHWORKS for Macintosh. It provides
VT102, VT320, and TTY (an older, basic terminal standard) emula-
tion (see figure 12.1). Compared to other third-party emulators,
MacTerminai doesn’t offer graphics capabilities, color, macros, or
redefinable keys. It provides basic emulation services. It, like most
other Macintosh terminal emulators, uses the Macintosh Com-
munications Toolbox (MCT). When purchased separately,
MacTerminai provides the basic MCT communication tools of
Serial and Modem. With PATHWORKS for Macintosh, additional
MCT tools are included that permit terminal emulation over LAT
and DECnet connections.
n
Macintosh DEC VAX
MacX
DECwindows is DEC’S implementation of the X- Window standard.
Macs equipped with MacX appear as DECwindows terminals (see
Chapter Twelve DEC VAX Connectivity
figure 12.2). MacX fully supports the DECwindow character sets. If
you’re considering using MacX, plan on picking up a three-button
ADB mouse, available from many third-part}' suppliers (the X-
Window system was designed for three-button mice), a minimum
of 8 M of RAM, and an Ethernet connection. MacX uses memory
and network bandwidth in large quantities.
Figure 12.2
MacX provides laiiMul
DECwindows emulation
|[omaMaciniosli.lntliis
example, MacX is using
DECneiasaiianspnii
overEilietneL
VAXshare File and Print Services
VAXshare is the name given to AFP file services and PAP print
services that run on a VAX. The VAXshare file server (see figure
12.3) can be set up to support multiple servers, each potentially
with multiple volumes. These volumes can be set up to reference
any VMS directory. These directories can contain VMS files (either
ASCII or binary) or foreign files, such as those from a PC. This way,
Mac users can edit VMS text files with Mac text editors or word
processors. They can also exchange binary application files with
corresponding VAX or PC applications.
Part Three Multivendor Netv^rorks
Figure 12.3
DiC'sVAXsIiafefile
server lurns a VAX into
an AfPiile server.
Matiniosli users are able
10 access VAX direciories
andliles.
Macintosh
DEC VAX
Figure 12.4
OneaspeciolVAXshare's
print services turns a VAX
into a PAP print spooler.
Maciniosh users are able
to send print jobs to ibe
VAX. Tbe VAX queues Ibe
job and sends inn the
"rearptinier.
The print services enable Mac users to spool LaserWriter print jobs
to the VAX and to print to DEC PostScript printers: they also enable
users of interactive VAX VT terminals to print to Apple LaserWriters
through a standard VMS print queue. These are diagrammed in
figures 12.4, 12.5, 12.6, and 12.7.
This program, or
virtual LaserWriter,
appears to the Mac
user as a real
LaserWriter.
This program, or
virtual Macintosh,
appears to the real
LaserWriter as a
real, printing, Mac.
Chapter Twelve DEC VAX Connectivity
This program, or
virtual LaserWriter,
appears to the Mac
user as a real
LaserWriter.
The PostScript file
is sent to a VMS
print queue and
then to a VMS
DECnet print driver.
Figure 12.5
Maciniosh users aie also
aMeiogriolioOEC
PosiScrim primers. Iliese
primers are eillter
iecilycortnetierlio
Elliernet or serially
coonecierlioaOiC
letotioal server.
9
Figure 12.6
MaciolosIrprimioploDEC
PosiScripi Primer.
DEC Ttrmlnel ServM
Terminal Server
Termiful Server
Program
Program
ASai
ASCU
Async j [
LAT
RS-232 /S
Ethernet
Figure 12.7
VMS users (Ilrai is,
rerorinal users) cao prim
loApplelaserlH/rirersas
illliey were DEC prioiers.
RS-232
Ethernet
Part Three Multivendor Networks
There are other similar AFP/PAP servers for VMS. Alisa Systems
offers AlisaShare (Alisa licensed portions of AlisaShare to Digital,
which eventually became VAXshare). Pacer Software has Pacer-
Share, which, like VAXshare and AlisaShare, provides AFP file
service and PAP print services.
With all these AFP and PAP services, the only supported transport
is AppleTalk for VMS. There is no support for the DECnet, LAT, or
TCP/IP transports.
DAL
Apple’s Data Access Language is a standard that provides a uni-
form relational database language. Apple offers DAL servers for
different platforms. Included with PATHWORKS for Macintosh is
Apple’s DAL server for VMS (see figure 12.8). When purchased
from DEC, it includes support for connections to Digital’s rela-
tional database, Rdb. If purchased directly from Apple, it also
supports Ingres, Oracle, Sybase, and Informix databases. (The
support for these additional databases can be purchased directly
from Apple.) Included are programming interfaces needed to write
Macintosh C or Pascal programs that will support the DAL service
and DAL-equipped HyperCard applications. The DAL VMS server
works with the AppleTalk for VMS transport and the normal serial
transport provided by VMS.
E-Mail
PATHWORKS for Macintosh includes two mail Interfaces. Mail for
Macintosh is a simple text-based mail application that uses a
DECnet connection, or an AppleTalk/DECnet gateway connection,
to VMS mail (see figure 12.9). AU-In-1 Mail for Macintosh is a more
ambitious application. It is an X.400-compliant client/server mail
application that supports binary enclosures, return receipts,
priorities, and other high-end mail features. (The adherence to the
X.400 messaging standard will make it easier to interface with
Chapter Twelve DEC VAX Connectivity
other foreign mail systems that also support the X.400 standard.)
DEC also offers companion All-In- 1 clients for DECwindows and
Windows-equipped PCs. All applications share a similar user-
interface.
r
Macintosh or PC Client VAX/VMS Host
Figure 12.8
Apple's DAI stiver ioi
l/MSisiPcyedwilli
PAIHWOHISIor
MaciPlosli.WilliilMac
users can access DEC'S
RilbrelatinnalilalaDase
witlianyDAlclieni
applicaiiun.
□
-M
□ □
Figure 12.9
Digital Pliers two mail
solutions with
PAIHWORKSIoiMac.
DotliuseDECnetontlie
VAX siile, as a transport.
Macs must either use the
DECnet transport or gu
through an Applelalk/
DECnet gateway.
Formats
Similar to the formats supported by the PC and UNIX computers,
the Mac and VAX are able to share ASCII text files and those
application formats that are binary-compatible.
Part Three Multivendor Networks
ASCII
The old standby, ASCII text, is an important common denominator
between the Mac and VAX. With VAXshare, Mac users can freely
move ASCII files back and forth. These ASCII files could be VAX
command procedures (DCL), VMS source code, PostScript files, or
IGES files from a VAX-based CAD/CAM system. VAXshare auto-
matically maps Macintosh creator and type codes to VMS files
based on their filename extensions. As an example, you might want
all your VMS text files that have a .TXT filename extension to be
automatically identified as Microsoft Word text files. This can be
done by editing a configuration file that maps extensions to Mac
creator and type codes:
Extension
Creator
Type
.TXT
MSWD
TEXT
In the above example, the .TXT extension is mapped to the cre-
ator code of Microsoft Word (MSWD). This code is assigned to
Microsoft and identifies all Microsoft Word files. The Type indi-
cates the specific kind of Microsoft Word file, in this case, TEXT. By
making this assignment on the VAX, any file with the .TXT filename
extension will be given these creator and file types (see figure
12.10). These are used to assign the proper Microsoft Word icon to
the file; they also tell the Finder to open the file with Microsoft
Word when you double-click the file’s icon.
Binary
Just as in the case with the PC and UNIX environments, more and
more applications are using the same binary format for their
document formats. These files can be seamlessly moved back and
forth between the two platforms with no conversion required.
Wlien conversion is required, there are several translation utilities
available on the VAX. Digital has an interchange standard known
Chapter Twelue DEC VAX Connectivity
as DDIF (see figure 12.1 1). DDIF supports a number of standard
Mac formats, such as MacPaint, PICT, Microsoft Word, and
MacWrite. Once these formats are converted (on the VAX) to the
DDIF format, they can be incorporated into DEC applications such
as DECwrite.
Macintosh
3T
DEC VAX
Figure 12.10
ASCII text files can lie
shaieillieiweenilieMac
andVAXenvironmenis.
Macintosh
AFP Client
Microsoft Word
Microsoft Word
Binary
AppleTalk
Ethernet
DEC VAX (Step 1}
AFP File Server
DDIF Converter
Microsoft Word
Binary
' AppleTalk for VMS '
DECnet
Ethernet
DEC VAX (Step 2)
DDIF Converter
DDIF
AppleTalk for VMS
DECnet
Ethernet
DEC VAX (Step 3)
DECwrite and
other DDIF apps
DDIF
' AppleTalk for VMS ^
DECnet
Ethernet
Figure 12.11
DECIias a common
intercliangefoimai called
DDIF; OIC also offers
(noi included wild
PAIHWDDKSIDDIF
conversion lonis for
MacWriie, PICI, MacPaini
and Microsoft Word.
Another approach to conversion is to use explicit conversion
utilities to convert one format to another. A good example of this
approach can be found in the Keypak line of translators from
Keyword Office Technologies (see figure 12.12). Keypak can be
used to convert a VAX word processing document format such as
WPS+ to a common Mac format such as Microsoft Word.
Part Three Multivendor Networks
Figure 12.12
Keypakconversioi
beiweenitieOECWP!l+
wordpiocessoianil
MatWrita.
Macintosh
AFP Client
Microsoft Word
Microsoft Word
Binary
AppleTalk
Ethernet
DEC VAX (Step 1)
AFP File Server
Keypak
Microsoft Word
Binary
' AppleTalk for VMS ^
DECnet
Ethernet
DEC VAX (Step 2)
Keypak
WPS+ Format
AppleTalk for VMS^
DECnet
Ethernet
DEC VAX (Step 3)
WPS+
Word Processor
WPS+ Format
'AppleTalk for VMS^
DECnet
Ethernet
Transports: AppleTalk for VMS,
DECnet, LAT
In the Macintosh/VAX world, the transports of each respective
platform are equally shared. The VAX speaks the AppleTalk proto-
col and the Mac speaks DECnet and LAT. As we’ll discover, the
choice of a transport for the most part is dependent on the desired
service.
AppleTalk for VMS
VAXes are able to speak the AppleTalk protocol because they’re
equipped with the AppleTalk for VMS (see figure 12.13). VAXes
equipped with AppleTalk for VMS appear on the network as
AppleTalk nodes. The services, VAXshare and DAL, appear as
AppleTalk sockets within the node. AppleTalk for VMS supports
AppleTalk routing, so VAXes with multiple Ethernet controllers can
be set up as AppleTalk Ethernet-to-Ethernet routers.
Chapter Twelve DEC VAX Connectivity
AFP, PAP and
DAL Services
Formats
AppleTalk
For VMS
Ethernet
Figure 12.13
Appletalk foi VMS equips
a VAX wiitiilie Appletalk
protocols. Appletalk is
usedbyoiosiolitie
PAtHWORKSIorMac
services.
DECnet
The most common VAX/VMS transport protocol is DECnet. VAX/
VMS systems usually come with DECnet, but Macs do not. So part
of the PATHWORKS for Macintosh solution is to install DECnet on
the Macintosh (see figure 12.14). DECnet on the Macintosh re-
quires a configuration process that includes the assignment of a
DECnet area and node number (similar to an AppleTalk network
and node number) and the generation of a list of other DECnet
nodes where communication is desired. This explicit address
assignment and node listing is contrary to the AppleTalk philoso-
phy of dynamic addressing and name binding using NBP.
Figure 12.14
DlCneilorMacinipsh
lurnsaMacInloaDiCnei
poile. A Mac so equipped
canbeauiooiaiically
backed up by a VAX no
thenelwork.
Despite the extra administrative work, there are reasons for install-
ing DECnet on your Mac. Once so equipped, the Macintosh hard
disk can be automatically backed up through DECnet copy com-
mands. The DECnet transport can also be used by MacX, the two
Part Three Multivendor Networks
mail applications, and terminal emulation. As we’ll see in the next
section, LAT cannot be routed and therefore will not provide wide-
area terminal services. An alternative is the CreflM protocol. Part
of DECnet, CTERM can be used as a routable terminal protocol.
PATHVVORKS for Macintosh includes a CTERM Communications
Tool (see figure 12.15) that can be used with MacTerminal, or any
other Macintosh Communication Toolbox (MCT)-compatible
terminal emulator.
Figure 12.15
CIEHMandOECnel
provide a roulable
iranspoiiproiocollot
letminal access in lire
DECVAXenvironmenl.
LAT
If your Macs are on the same Ethernet as the VAX, you’ll be able to
use the LAT MCT tool to directly connect to the VAX. Conventional
VT terminals, or Macs running terminal emulators over serial lines,
aren’t capable of speaking LAT over Ethernet; instead, they speak
the As>mc protocol over RS-232. A device called a terminal server,
or DECserver, is used to make the connecdon between multiple
terminals and any LAT-speaking VAX (see figure 12.16). By speak-
ing LAT direcdy, as shown in figure 12.17, the Macintosh can
connect direcdy and avoid the intermediate service of the terminal
server.
As mentioned before, LAT cannot be routed. It was designed for
use on a LAN. LAT is a time-critical protocol, so care must be taken
when LAT is used on bridged Ethernet. Even though a bridged
Ethernet logically appears as a single Ethernet LAN, if the bridge
connection introduces a significant delay, LAT may have problems
Chapter Twelve DEC VAX Connectivity
maintaining connections. In these cases, and when a routed
network exists, the DECnet CTERM protocol mentioned before
might be a better choice.
Figure 12.16
lnlfieolililays,VAK
tenninal users relied on
terminal servers 10
connect serial terminals
tn networked VAXes.
Macintosh
VAX/VMS
Figure 12.17
Macs on [tliernet that are
equipped with the lAI
tool bypass the terminal
server and can access
any lAT host |i.e.D(C
VAX) on Ihe Ethernet lAK.
AppleTalk/DECnet Gateway
The AppleTalk/DECnet Gateway converts AppleTalk protocols
into the corresponding DECnet protocols. A prime example of
its use is with the DECnet-based PATHWORKS mail services
Part Three Multivendor Networks
Figure 12.18
IheApplelalk/DECnei
Gateway conveits
AppleTalk proipcols into
OECneipioiPcolsaPiivjce
versa. At the ntomepi.
it's main use is to let
Macs access DECnel
mail services witliouttlie
ailminisirative overhead
oiputtinp DECnel on
(see figure 12.18). Because the VAX mail services are only acces-
sible through DECnet, there are two ways a client Mac can con-
nect: by putting DECnet on the Mac, and by using AppleTalk and
going through the gateway. The gateway is selected through the
AppleTalk/DECnet MCT tool from within the mail applications.
AppleTalk/LAT Gateway
The LAT protocol is restricted to Ethernet. How do LocalTalk Macs
connect to terminal services? They can’t use LAT directly because
it only works over a direct Ethernet connection. You could use the
DECnet CTERM tool and a LocalTalk/ Ethernet router that routes
DECnet, but you would have to put DECnet on your Mac and put
up with the extra administrative overhead associated with DECnet.
Fortunately, Apple and DEC came up with another alternative.
The AppleTalk/LAT Gateway runs on a Mac that has LocalTalk
and Ethernet connections (see figure 12.19). This effectively turns
the Macintosh into a terminal server. The Macs on the LocalTalk
side speak the AppleTalk protocol to the server; the server speaks
Chapter Twelve DEC VAX Connectivity
LAT to the VAX hosts on the Ethernet. Figure 12.20 shows various
screens from the AppleTalk/LAT Gateway as well as the client
settings of the AppleTalk/LAT Tool.
The gateway is also useful for dial-up users who attach to the
network with Apple's Remote Access software (ARA). As shown in
figure 12.21, incoming ARA Macs speak AppleTalk. By going
through an AppleTalk/LAT Gateway, a remote Mac user can
engage in terminal emulation with the VAX, whUe retaining the
capability to access AppleShare file servers and LaserWriter print
services.
Figure 12.19
Macsconneciedon
locallalk can access lAl
leiminal services by
goinoihiDuobihe
Applelalk/lAl Gateway.
Figure 12.20
IheApplelalk/lAI
Gateway runs on a
llllacintosh. Client Macs
usetheApplelalk/lAl
tool to access the
gateway.
Part Three Multivendor Networks
Figure 12.21
IlieApplelalk/DECnel
Galeway lets reiooie
useiseegageieteiinipal
emplaiionatihesanie
lime as Ollier
Applelalk services.
Cabling: LocalTalk and Ethernet
The cabling choices for Macs in a VAX environment are primarily
LocalTalk and Ethernet. Since it’s unlikely that VAXes will ever
sport LocalTalk interfaces, the options are few: either put Ethernet
on the Macs, or use some sort of LocalTalk/ Ethernet router.
LocalTalk
The cabling choices for PATHWORKS-equipped Macs are primarily
LocalTalk and Ethernet. Because VAX computers don’t come with
LocalTalk interfaces, some sort of LocalTalk/Ethernet router will
be needed (see figure 12.22). The choice of a LocalTalk router
depends on the desired services and protocols. If you plan to use
DECnet, then a multiprotocol router capable of routing DECnet
will be needed. If you plan to stick to the AppleTalk services, or
plan to access the DECnet services (such as E-mail) through
AppleTalk, then any generic AppleTalk LocalTalk/Ethernet router
will do.
Because most VAX sites have extensive twisted-pair wiring that’s
used for terminal access, you might want to adapt this wiring to the
phone- type LocalTalk connectors, such as Farallon’s PhoneNET.
Chapter Twelve DEC VAX Connectivity
Figure 12.22
Macs with ilieOECnei
iianspoft cat) still use
locallalkwhena
iDcallalkOECneiiouiei
is used.
Ethernet
If your VAX network is Ethernet-based, which most are, then it’s
likely that Ethernet-equipped Macs can be easily incorporated into
your network. Digital offers a number of desktop Ethernet prod-
ucts that can be used to connect PCs and Macs to the network.
Digital recently announced a line of multiprotocol routers with
support for the AppleTalk protocol.
Serial RS-232/Diai-Up
With MacTerminal, Macs can take the place of VT terminals on
Ethernet or serial connections. These connections can be made
directly to the VAX or through a terminal server.
With serial connections, remote MacTerminal users can use
modems and log onto the VAX, but this connection only provides
terminal services. Often, remote Mac/VAX users need to access
AppleTalk services in addition to terminal services. Apple and DEC
don’t offer a solution, but Computer Methods Corporation has a
product called AsyncServeR that essentially turns the VAX into an
AppleTalk Remote Access ser\'er (see figure 12.23). Most VAX sites
have banks of dial-up modems already in place; AsyncServeR takes
advantage of these resources by enabling any Mac with the Remote
Access software to dial in and access not only the AppleTalk
services on the VAX, but also the services on the entire AppleTalk
internet.
Part Three Multivendor Networks
Figure 12.23
AsyncSeivelUioiii
Compiiiei Meihods. uses
ilieApplelalklorVMS
enviionmeniasa
lounila(ionloraVA!l-
basedApplelalk Remote
Access Seiver.
n
y y— ^ — T
Remote Macintosh
Macintosh,
LaserWriters and other
AppleTalk>based
Services
Conclusion
Digital redefined computing in the 1970s; Apple did the same in
tire l980s. In the 1990s, Apple and Digital both realize that the
network is the common battleground for computing services.
Digital, with its PATHWORKS for Macintosh product, has recog-
nized the importance of the Macintosh. They have successfully
integrated two supposedly closed architectures by providing
terminal emulation, DECwindows (X-Window), AFP file and print
services, database access, and two E-mail packages.
IBM Connectivity
till a large part of the corporate world, IBM’s
mainframe and minicomputers are an important
part of Apple’s network strategy. Once again, the
Macintosh proves its worth as the “universal
client,” easily connecting to the Big Blue world.
Services: Mainframe and AS/400
The IBM computing world (PCs excluded) is comprised of the large
mainframe computers, such as the System 370 (S/370) and 390
(S/390), small mainframes such as the ES/9000, and midrange
systems like the AS/400. These systems aren’t exactly showcases
of client/server technologies, as over the years the typical user-
interface has been through terminal emulation.
Terminal Services
In the past, IBM mainframe access was typically made with the
3270 family of terminals. Today, 3270 terminal emulation is
commonplace on PCs and Macs (see figure 13.1). As with VT-type
terminals, the sales of so-called “dumb” terminals are declining
while the use of intelligent workstations is increasing.
Part Three Multivendor Networks
Figure 13.1
AMaciniosliminoa
3210eniulaiofovera
coaxial connectioa IQ
Ihehosi.
There are a number of 3270 terminal emulators available for the
Macintosh. These emulators enable suitably-equipped Macs to
access IBM mainframes. Most Mac emulators add value by adding
features such as programmable “hot keys” or macro capabilities.
Some emulators support keyboard remapping, where 3270 keys
can be reassigned to the Macintosh keyboard. A common feature
in 3270 emulators is the ability to perform IND$FILE file transfers
from within the terminal session. Four popular 3270 emulators
include Apple’s SNA»ps 3270, DCA's IRMA Workstation for Macin-
tosh, ASC’s asc3270, and Avatar’s MacMainFrame.
The 3270 standard is used by the large mainframes, while the
midrange AS/400 uses the 5250 standard. 5250 emulation on the
Mac is not yet as common as emulation of the older 3270. 5250
emulators offer similar features as their 3270 counterparts and are
offered by Andrew and ASC (see figure 13.2). Conventional termi-
nal screens can be a bit daunting: to make things a bit easier, IBM
terminal sessions can also be front-ended with user-ft’iendly
interfaces that interact with the terminal data stream.
Chapter Thirteen IBM Connectivity
Macintosh IBM AS/400
One of the most popular IBM mainframe applications is the PROFS
mail environment. Normally, PROFS is accessed with a conven-
tional terminal. However, there are several Mac terminal front
ends that put a Mac interface onto PROFS (see figure 13.3). Two of
these — MitemVision from Mitem and Executive Workstation from
MediaWorks — are HyperCard-based.
IBM Host
Macintosh i
(Cluster Controller)
MitemVision
HyperCard
PROFS
Mail Environment
EBCDIC
EBCDIC
SNA
I [ SNA
Coax
(Apple or third party)
^ y Coax i
Figure 13.2
AMacinloshiunninoa
5250emulaloro«era
iiiaxialconneciionio
the host
Figure 13.3
PROFS access wilii
MilemVisioa.a
HtpeiCatil'Iiaseil
from end.
Part Three Multivendor Networks
HyperCard can also be used to create custom terminal front ends.
With products from Avatar, DCA, and Mitem, developers are able
to create HyperCard applications using custom XCMDs and XFCNs
that provide controlled access to the terminal data stream. Other
front-ending environments use their own development environ-
ments to make quick work of customized interfaces. These include
Blacksmith from CEL, SimMac from Simware, and Both from
Connectivite.
Database
Apple's DAL server for IBM provides DAL clients access to IBM
DB2 and SQL/DS relational databases (see figure 13.4). Connec-
tions to the DAL server can be made through a 3270 terminal
connection, serial links, or with IBM’s network transport, SNA.
As mentioned elsewhere in this book, DAL clients can be either
Mac- or PC-based. They can be developed from scratch with
conventional Mac programming environments or authored with
HyperCard or Serius. DAL clients can also be used out of the box,
since many Macintosh applications, such as spreadsheets and
databases, offer built-in DAL support.
Figure 13.4
Apple's OAL Server for
IBM liDSis provides
MaciitloshlandPCIOAL
clienis access 10 IBM
relational daiabases
sucliasflBl.
Chapter Thirteen IBM Connectivity
Formats: Mainframe and AS/400
Although the Mac and IBM hosts speak different fundamental
languages (ASCII versus EBCDIC) they still enjoy a high degree of
interoperability. This is due to the use of standardized terminal
formats (as in the case of 3270 terminal emulations) and format
translators that provide ASCII to EBCDIC translations in real time.
EeCDIC
The IBM PC was their first computer to use ASCII. Before that, IBM
mainframes used EBCDIC (Extended Binary Coded Decimal
Interchange Code) which, in a technique similar to that of the
ASCII code, uses an eight-bit code to represent characters. Conver-
sions between EBCDIC and ASCII are typically handled by protocol
converters that can either be hardware- based (see figure 13.5) or
software-based.
Figure 13.5
lOM'sIBCDIC codes aie
itansloiedinioMaciniosli
ASCII codes Willi ilie IBM
7171 Piolocolcomieiier.
Transports: Mainframe and AS/400
Traditionally, the IBM transports have been unique to the IBM
world, focusing on terminal-based applications. During the past
several years, as IBM has pursued a peer-to-peer environment,
tlie rest of the industry' has started to adopt IBM’s new standards.
Today, the Macintosh supports all of the major IBM transport
Part Three Multivendor Networks
Figure 13.6
Apple's SNA*psGaiewBV
tonnecisApplelalk-
conveisaniMacsioan
IBMhesi.
protocols, in addition to the IBM-supported industry standards
such as TCP/IP.
SAA, SNA, and APPC
SNA stands for System Network Architecture and is IBM’s primary
transport protocol. When a Macintosh runs an IBM terminal
emulator, it’s relying on a portion of the SNA protocols. SNA is only
one part of IBM’s networking strategy, known as System Applica-
tion Architecture (SAA). SAA defines an environment that supports
peer-to-peer applicadon services, which is a departure from the
traditional host-based IBM world.
Part of IBM’s SAA/SNA plans include APPC. Standing for Advanced
Prograrn-to-Program Communications, APPC uses a protocol
called LU 6.2 (the LU stands for Logical Unit). Similar enhance-
ments to SNA networking are part of IBM’s Advanced Peer-to-Peer
Networking (APPN) protocol. APPC and APPN are still relatively
new, but compatible applications are starting to appear. Apple’s
SNA'ps gateway is one example (see figure 13.6). It supports both
the APPC and APPN protocols and will undoubtedly be used as a
development platform for future peer-to-peer applications that
work in both the Macintosh and IBM environments.
Chapter Thirteen IBM Connectivity
TCP/IP
Many companies, in an attempt to support multi-platform com-
munications, are using the TCP/IP protocol in place of IBM’s SNA
protocol. TCP/IP provides a common transport protocol that,
unlike SNA, is supported on Macs, VAXes, UNIX, and many other
platforms. Support for TCP/IP services is beginning to grow as
companies such as ASC begin to offer Macintosh terminal emula-
tors that use MacTCP to communicate with similarly equipped
IBM hosts (see figure 13.7).
Macintosh
IBM AS/400
5250 Terminal
Emulation
5250
Applications
EBCDIC
\ / EBCDIC
TCP/IP
] [ TCP/IP
» Ethernet
^ y Ethernet <
Rgure 13.7
5250le[minaleniulalioii
Willi asc525D over a ICP/
iPneiwoikconnectiim.
Media: Mainframe and AS/400
There are two primary ways that physical connections are made in
the IBM environment. One way is through direct connections,
which are primarily used for terminal sessions. These direct
connections can be made over coaxial cabling or through a Syn-
chronous Data Link Control (SDLC) connection. The other way
connections are made is through a LAN, which can be used for a
wide variety of services. Token Ring is currently the predominant
LAN cabling system in the IBM world, but Ethernet is slowly
gaining in acceptance.
Part Three Multivendor Networks
Direct Connections
The most common direct connection to an IBM mainframe is
made with a coaxial (coax) cable. Traditionally, this has been the
method employed by IBM terminals for years; it's also used by
Macs and PCs running terminal emulation programs. Because
Macs don’t come with a coax connection, an expansion card is
required. The coax card is usually connected to an intermediate
device known as a cluster controller, which is used to connect
multiple coax devices to the mainframe. As shown in figure 13.8,
terminals directly connected to the IBM AS/400 midrange com-
puter use a variation of coaxial cable called twinaxial cable. Unlike
coaxial cable, which has a single conductor surrounded by a
second shielding conductor, twinaxial cable has two conductors
inside. Macintosh coax and twinax cards are available from Apple,
Avatar, and DCA.
Figure 13.8
AMacimosliconnecieilio
an IBM AS/4011 host wilh
aiwinaiialcable.
Macintosh IBM AS/400
SDLC connections are made with high-speed synchronous mo-
dems. As shown in Figure 13.9, an SDLC card is used to connect
the Mac to the modem, which in turn connects to other modems
Chapter Thirteen IBM Connectivity
on the mainframe. Apple’s Serial NB card can be used to make
an SDLC connection. Avatar also offers cards that perform the
function.
Figure 13.9
A Maciniosh connected
loanlBMIiosiwi
synclifonous modems and
a SDlt connection.
Both the coax and SDLC connections tend to be expensive, either
because they require a number of modems (in the case of SDLC) or
because the coax connections need available ports on the cluster
controller. When ports on the cluster controller are no longer
available, another controller is required. To make a number of
cost-effective connections to the mainframe, a gateway is often tlie
best option.
Avatar and DCA have software that enable a Macintosh to act as a
SNA gateway when coax connections are used. A Mac that acts as
a SNA gateway has one coax connection to the cluster controller
and another LAN connection to the participating Macs. These
Macs run the appropriate vendor’s client software and can be on
LocalTalk or Ethernet networks. A similar gateway is also used
for SDLC connections. A Mac acts a gateway by connecting a
LocalTalk or Ethernet LAN to a single, shared SDLC connection
(see figure 13.10). An example of this type of gateway can be found
in Apple’s SNA*ps gateway. In addition to Macs that act as gate-
ways, there are also several dedicated gateway boxes drat perform
the same function.
Part Three Multivendor Networks
Hgure 13.10
AMacinloshconnecieilio
anIQMhositaolian
SNA oaieway Willi an
SDlClink.
Token Ring
Of course, Token Ring is the leading LAN technology used in the
IBM environment. There are number of companies (Apple, Avatar,
and DCL\) that make Token Ring cards for the Mac. Macs using
Token Ring can be attached directly to the mainframe (see figure
13.11), or to a gateway.
Figure 13.11
AMacinloshconnecieilio
anlBMhosiihiougha
direci Token Ring link.
EBCDIC
IBM Host
Macintosh i
(Cluster Controller)
Client
Applications
Host
Applications
EBCDIC
SNA
SNA
Token Ring
(Apple or third party)
Token Ring
Ethernet
Ethernet is gaining in popularity in the IBM world, particularly as
TCP/IP continues to become more prevalent on IBM mainframes
(see figure 13.12). The most popular means of Macintosh Ethernet
connections is through a gateway.
Chapter Thirteen IBM Connectivity
Macintosh
IBM AS/400 Host
Figure 13.12
AMacintosticDnnecieiMD
anlBMIiosltaolia
diiecililiernellink.
Conclusion
The IBM mainframe world is changing, but it is still typified by
terminal-type applications. The Macintosh can easily adapt to this
environment with numerous terminal emulators and front-ending
programs. These products support IBM’s cabling (coax, twinax, or
Token Ring) and protocols (SAA, SNA, APPC, and APPN). They also
support non-traditional cabling and protocols such as LocalTalk,
Ethernet, TCP/IP, and AppleTalk through the use of gateways.
gn, implementation
“Four”
Part IV brings everything together and outlines the process of design,
) > - j
implementation, and management. First, in Chapter 14, dw practical issues
i 0 1^0 1 1 Ifc r 3* iCi i) i •( t < * ' > f; •' -;J '■
centered around network implementation, such as wiring and maintenance,
are examined; the NetPICT technique is used to solve specific networking
] n ! olt'n h'?i " In : : o
I
problems. Once a network is in place, the focus shifts to other
i n ' ; Vi , ; MT; r'Ol ?
issues — discussed in Chapter 15 — such as keeping track of users and
services, troubleshooting, performance monitoring, and security.
Network Design,
Implementation,
and Management
Design and
Implementation
his chapter covers the basics of network design
and implementation. Several common
Macintosh network scenarios will be discussed
and analyzed. These networking scenarios
should make it easy for you to identify your
current (or planned) network environment. Also
included in this chapter are tips and hints that will
streamline your networking tasks.
Top-Down Design Techniques
The NetPICT diagrams used throughout this book conveniently
break up networking into four layers. This “divide and conquer"
approach, inherent in the layering process, makes it easy to attack
networking problems. This is similar to the process in which
software projects, or computer programs, are solved by breaking
the task into smaller “digestible” components.
Combined, the four layers represent a particular solution to the
overall networking problem. However, each separate layer can —
and should — be addressed on its own merits.
Part Four Network Design, Implementation, and Management
The process should start at the top. Simply put, this means asking,
"What Services are required to do the job?” This should be the
driving force behind any network plan. Unfortunately, this is not
always the case. All too often, companies start the process from the
bottom up. They try to dictate Cabling choices. Transport proto-
cols, and particularly computing hardware in an attempt to
achieve some measure of standardization. By starting at the
Semce layer, it is much easier to stay focused on the users’ needs.
Once the question regarding Services is asked, there are a number
of other questions that naturally follow. "Do these Services provide
for growth? Are we going to be locked into a specific vendor’s
products?”
Organizations must begin to look at network Services as strategic
decisions that can significantly affect their operations. These
Services can be as mundane as shared printing to a laser printer, or
as pervasive as an enterprise-wide E-mail system. Let’s use E-mail
as an example.
Let’s say that the Ajax company decides tliat employee communi-
cation is a crucial aspect of their business. They decide to imple-
ment a corporate-wide E-mail system to facilitate this communica-
tion. They decide on an X.400-compatible mail application. One
reason that they made this decision was because of the portability
of the X.400 standard. They reasoned that widespread support of
the X.400 messaging format would enhance tlieir options for future
expansion and grov\'th. All in all, a reasonable approach.
Once the Ajax company made the decision to go with the X.400
standard, the rest was comparatively easy. A suitable Transport
and Cabling decision is left as the remaining task. It should be
obvious that while the decisions of a Transport and Cabling are
important, they are secondary to the choice of a Service and
Format.
Chapter Fourteen Design and Implementation
While it’s hard to simplify and generalize all networking situations,
here are a few suggestions that may help to organize and prioritize
your networking plans:
1 . Establish and evaluate the required Services. Do they meet
your requirements? Are they supportable?
2 . During this process, be aware of the limitations and restric-
tions imposed by the Formats used by the Services you have
selected. Will your data be held hostage by the application
vendor, or can it be transported to other applications and
Services?
3 . Once you've chosen the Services, along with their attendant
Formats, start thinking about a Transport protocol. Because
Services are often protocol-specific, the choice of a Transport
protocol may be restricted.
4 . While selecting a Transport protocol, familiarize yourself with
the administrative requirements inherent in the protocol.
5 . Is the candidate Transport protocol supported on likely
Cabling mediums?
The preceding five suggestions may be somewhat simplistic, but
they help to illustrate and represent the general process of working
from the top down during the network design process.
Using NetPICTs to Help with
Systems Integration
This book comes with a disk that contains numerous NetPICT
symbols. Many of the symbols are duplicated from products
Part four Network Design, Implementation, and Management
discussed in the book; others depict other common networking
products.
The disk contains a viewer application called PICTviewer. It is used
to view the NetPICT symbols. For each NetPICT symbol there are
two files: a PICT file that contains the graphical image, and a text
file that contains a textual description of the image. The text file
has the same name of the PICT, but with the addition of a “.txt” file
extension. For example, the PICT file portion of Apple’s Internet
Router is named “Apple Internet Router,” while the text file
portion is named “Apple Internet Router.txt.”
The NetPICT viewer provides an easy means for you to view and
read informative data on hundreds of products. Of course, since
the symbols are in the PICT format, they can be easily incorporated
into documents made by your favorite drawing program. The same
can be said for the text (.txt) files. These text files can be added to
your drawings to annotate the figures. Many drawing programs,
such as MacDraw Pro, can import text files directly. If your drawing
program can’t directly import text files, then you’ll have to copy
and paste with a text editor or word processor. By using a drawing
program, you’ll be able to import those NetPICT symbols that
represent the fundamental parts of your present or planned
network. You’ll be able to see where bridges, routers, gateways, or
even format translators are required.
To use the NetPICT viewer, you’ll need HyperCard 2.1 (or the
HyperCard viewer). You might also want to copy the NetPICT
viewer, along with the symbol folder, to a hard disk.
1 . Launch the NetPICT viewer by double-clicking its icon.
2 . Click several times to close the splash screen.
Chapter Fourteen Design and Implementation
3. Use the pop-up folder menu to browse through the NetPICT
symbol folder hierarchy.
4. Once you’ve found the appropriate symbol, click once on the
name that appears in the window directly beneath the pop-
up menu.
The chosen NetPICT symbol will appear in a separate win-
dow to the right of the viewer screen. You can reposition
this window at any time. You can even enlarge it to the
size of your monitor by clicking on the Zoom box located
at the upper right hand corner of the symbol window (see
figure 14.1).
Figure 14.1
Hieimaoe window of the
NeiPICI viewer can be
repositioned as needed In
folly view tbe NeiPICI
synibol.
5. To restore the symbol window to the right side of the viewer
window, click once on the magnet icon. This will automati-
cally dock the symbol window to its default position at the
right of the viewer window (see figure 14.2).
Part Four Network Design, Implementation, and Management
Figure 14.2
Hiemagnel icon re-docks
lire PICT image at lire
upper right corner of tlie
viewer application
window.
Wiring Strategies
As mentioned throughout this book, the Macintosh supports a
number of cabling systems, from LocalTalk to FDDI. We’ve dis-
cussed the pros and cons of each cabling system, but there are
numerous issues involved with the design and implementation of
any wiring system. While it’s hard to generalize and cover all
possible networking situations, here are a few things to keep in
mind.
• Don’t get locked into a specific vendor’s proprietary compo-
nents or cabling scheme. Stay with the accepted industry
standards.
NOTE: One exception is Apple’s Ethernet system. It uses a special
connector to link the device with the external attachment unit. Since the
network side of the attachment unit (either 10Base-5, 10Base-2, or
lOBase-T) stili conforms to industry standards, it’s less important that
the device side is somewhat non-standard. To their credit, Apple has
published and made this standard available for any vendor to use. The
Apple Ethernet system has already been adopted by several Macintosh
networking vendors. Perhaps other non-Macintosh networking
companies will follow suit.
Chapter Fourteen Design and Implementation
• Don’t limit your network and wiring plans just to data.
Consider integrating your network plans with other forms of
data, such as voice and video. You may be able to save money
by installing these different cabling systems at the same time.
In many cases, the different systems can use the same face-
plate.
• Plan for the future. You may be planning to use only Local-
Talk and Ethernet, but you should seriously consider the
installation of twisted-pair wiring and fiber backbones that
are capable of supporting current and future high-end
cabling systems.
Example 1: Four-Pair Twisted Pair .
One such industry standard involves the use of four pairs (10Base-T
Ethernet requires two pairs) of Unshielded Twisted-Pair (UTP) wiring.
The most encompassing standard for UTP is published by the Elec-
tronic Industries Association and the Telecommunications Industry
Association (EIA/TIA). The standard is commonly referred to as EIA/TIA
568 and TSB-36, which sets a standard for “Category 5” UTP media.
Category 5 UTP anticipates the upcoming standards that will support
bandwidths of 100 Mbps (Category 4 UTP supports the conventional
Ethernet bandwidth of 10 Mbps). These standards include copper
versions of the FDDI standard (CDDl) and 100 Mbps implementations
of Ethernet. This cabling is also sometimes referred to as “Level 5"
cabling.
Example 2: Fiber-Optics
Another example of planning for the future is the choice of fiber-optic
cabling. Fiber optic cabling is becoming more commonplace as a
medium for Ethernet backbones. In fact, a new emerging Ethernet fiber
standard, IEEE 802.3 — lOBase-F — will be published shortly. Fiber can
be run over great distances and is immune from most forms of
electromagnetic interference (just don’t get it too close to a black hole).
Fiber optic cabling is also the basis for the FDDI standard. Fortunately,
the two standards overlap somewhat and it’s desirable to implement
Part Four Network Design, Implementation, and Management
fiber networks that conform to the FDDI standard (ANSI X3T9.5: FDDI)
as well as the current and planned Ethernet fiber standards. The moral
of this story is to plan for FDDI even though you're only going to use
lowly Ethernet. You'll appreciate this approach in the future when FDDI
cards sell at CompUSA for $129.
• If it appears that your LAN will need to support more than 50
users, plan on implementing a structured wiring approach to
your network design. It’s money well spent. By implemen-
ting such a system, the ongoing costs involved in support,
troubleshooting, expansion, and equipment relocation will
be minimized.
• In accordance witli your budget, plan on running multiple
cable runs to each office or cubicle. The incremental cost of
the additional cabling is relatively small when compared to
the minor increase in labor cost. Also, don’t scrimp on the
number of jacks provided to an office or cubicle. You can
choose faceplates that have a single, double, or quadruple
complement of connectors. While it’s often possible to daisy-
chain within a room, don’t plan on this as a part of the design.
Consider that you might want to add printers or other
networkable peripherals to a room at any time in the future.
You might also need to support devices that require multiple
connections to the LAN, such as workstations, routers, and
gateways. A good rule of thumb is to plan for two times as
many connections as you actually require at present.
• Make sure that fiber and copper cable runs are within the
specified length limitations, by carefully planning the loca-
tion and distribution of wiring closets. Be sure to allocate
additional cable length for vertical distances (not usually
shown on floor plans) and the routing losses that occur when
cable is snaked above ceilings and through walls.
Chapter Fourteen Design and Implementation
• Speaking of wiring closets, plan the layout and design of these
rooms carefully. Make sure that the room is well lighted,
ventilated, and capable of being securely locked. If the room
is large enough, use industry standard 19" racks to contain
the network devices. Obviously, since the room may house a
number of routers, gateways, and other network devices,
adequate electrical service is crucial. If you have the room,
you may also find it convenient to install a small desk with a
Mac that can be used for troubleshooting and maintenance.
This Mac can also be used to maintain network configuration
information, wiring diagrams, and problem logs.
• Are your cables routed in such a manner as to avoid electro-
magnetic interference and undue mechanical stress? For
example, don’t drape cables around fluorescent light fixtures,
and don’t hang wet laundry from cables strung from the
ceiling. Use common sense and consult the appropriate local
codes and requirements.
• Plan to have your wiring tested before use. This is a particu-
larly good idea for large, expensive installations that use
structured wiring. The time to discover breaks or discon-
tinuities in the cable is be/ore network deployment. If you use
a wiring contractor, be sure to include testing and validation
as part of the agreement. The devices used for testing can be
rather expensive, so it’s likely that this task is better handled
by the cable installers.
• Adopt an adequate cable marking and identification scheme.
Each and every cable should be unambiguously marked — in a
location that will be visible after installation. Be sure to mark
both ends of the cable. Make sure the marking is permanent
and won’t easily be erased or detached. There are many
adequate marking systems. Some utilize heat-shrinkable
tubing, while others use labeled wiring ties.
Part Four Network Design, Implementation, and Management
• Document everything from the outset. Create a logical and
physical diagram of your network. If you’re using a cabling
system that has physical hardware addresses, such as
Ethernet, register these numbers along v\ath their location
and description. For this task, a spreadsheet or a simple
database (such as FileMaker Pro) is ideal.
• Call, without delay, the Black Box Corporation at (800) 552-
6816. They’ll send you a free catalog that’s chock full of many
networking products. It’s like the “Sears Catalog” for
networkers.
Using diese twelve suggestions as a guide, let’s now examine some
common networking scenarios. These scenarios will serve severed
purposes. First, they will expose you to a number of different
wiring schemes, along with their benefits and disadvantages. Next,
because the scenarios progress from the simplest to the most
complex, you’ll be able to establish a growth path for your environ-
ment.
Scenario la: Single LocalTalk Network
Using a Daisy-Chain Topology
The simplest Macintosh network is the single LocalTalk daisy-
chain network (see figure 14.3). This network has been around
since 1985 and in many cases is still an entirely acceptable network
architecture. This network is inexpensive. All that is required is the
LocalTalk (or phone-type) connectors and the necessary wiring. In
nearly all cases, the phone-type connectors (Farallon's PhoneNET)
are recommended. Because these connectors use the telephone-
style RJ-11 jacks, you’ll be able to make your own cables by pur-
chasing a spool of twisted-pair phone wire, jacks, and a crimping
tool. Electronics stores such as Radio Shack should have everything
you’ll need.
Chapter Fourteen Design and Implementation
Figure 14.3
A single iDcallalk
neiwoik Willi daisif-cliain
logology.
The cost of this network is the lowest of any of the other scenarios.
Expect to pay $25 for each LocalTalk/phone-type connector. With
the additional cost of the wiring, the total per-node connection
cost should be well under $50.
With this network design, you’ll be able to provide modest band-
widtli for any number between two and thirty devices. The maxi-
mum distance of the network may vary somewhat, but plan on a
maximum of 1,000 total feet of wiring. If you need more distance,
consider the use of a LocalTalk repeater.
The tradeoffs for this design are the low cost and ease-of-
installation versus the limited bandwidth of LocalTalk. If you plan
on using services that demand high bandwidth, then the simple
LocalTalk is not for you. Applications that demand high bandwidth
include networked databases, demanding AFP services where large
files are frequently transferred, and demanding print jobs submit-
ted from a number of users. As a rule of thumb, try to avoid put-
ting more than five networked printers on a single segment of
LocalTalk. If you need this many printers to service your users,
then a routed scenario or a higher bandwidth cabling system is
probably required.
This network scenario is ideal for small offices that deal v\ith
smaller documents and have modest print requirements. It also
helps if the physical layout of the area to be networked can easily
support the daisy-chain topology. If the devices are all located
within a single room, then the daisy-chain can simply run around
Part Four Network Design, Implementation, and Management
the periphery of the room. On the other hand, if your network
nodes are distributed throughout a large building that doesn’t
provide ready wiring access, then a more structured wiring ap-
proach might make more sense.
One point to remember about the Lx)calTalk daisy-chain is that
devices can be added to the ends of the chain at any time, but if
you need to add a device somewhere in the middle, the network
will have to be disconnected to attach the new LocalTalk connec-
tor. This addition won’t take long, but the users will have to put up
with a slight interrupdon in network services.
These considerations can be summed up as follows:
+ Ideal for small workgroups (2-20 devices) in close proximity
+ Easy to wire
- Limited bandwiddi and throughput
- Network disrupted when devices are added
- Limited to approximately 30 devices
• Maximum cabling length of 1,000 feet
Scenario 1b: Single Ethernet Network
Using a Daisy-Chain Topology
If you find the simplicity of Scenario la appealing, but you’re
concerned about the limited bandwidth of LocalTalk, then an
Ethernet solution may be just the ticket. In this case, we’ll use the
thinwire variant of Ethernet known as lOBase-2 (see figure 14.4).
Although twisted-pair Ethernet (lOBase-T) is getting all the atten-
tion in the press, a thinwire daisy-chain is still a valid solution for
small network installations in one compact area.
Chapter Fourteen Design and Implementation
Rgure 14.4
Asinileliheineineiiiiiiiili
using diinwiiellOBase-I)
catilinoaniladaisif'tliain
lonology.
With the thinwire tee connectors, the nodes simply can be con-
nected one after the other. The total cable length is 185 meters, or
about 600 feet. Like the LocalTalk example in Scenario la, changes
made to the network will cause a brief disruption of services.
Apple’s thinwire transceivers have a unique self-terminating
feature that engages the terminating resistor whenever the connec-
tion is broken. With these devices, a break in the middle of a
segment will not cause the two segments to fail. Each segment will
continue to operate on its own. If you use the conventional
thinwire devices, be sure that each free end of the cable has the
terminating resistive cap in place.
This scenario is ideal for small workgroups that require more
bandwidth than LocalTalk. The Ethernet cabling should provide a
300 to 500 percent improvement in throughput over the LocalTalk
cabling. The number of nodes is still somewhat limited, as 30 is the
recommended maximum number of devices on a single segment
of thinwire.
One difficulty with thinwire is the cabling. Each cable must have
the twist-lock BNC connector fittings. You can purchase pre-
assembled lengths of these cables, or invest in the bulk fittings and
cable and crimp your own. You might need to go to a networking
supply house, such as Black Box, to find these components and the
proper crimping tools.
The cost of a thinwire network is somewhat more than a LocalTalk
network. If your Macs, LaserWriters and other network devices
aren’t equipped with Ethernet, then Ethernet cards will be
Part Four Network Design, Implementation, and Management
required. Expect to pay $200 to $300 per card. Devices that don’t
offer Ethernet connections (such as PowerBooks and certain
LaserWriters) will require special Ethernet devices that connect via
the SCSI port or through the LocalTalk port. Add the cost of the
cables, and you’re likely to spend $300 to $400 per node.
These considerations can be summed up as follows:
+ Ideal for small workgroups in close proximity
+ Easy to wire
+ Improved throughput compared to LocalTalk
- Network disrupted when devices are added
- Segment limited to approximately 30 devices
• Maximum cabling length of about 600 feet
A similar scenario to a single Ethernet network involves the use of
FDDl as a cabling medium. Of course, the per-node connection
cost is considerably higher. If your network requirements are
severe, then this approach should be considered. Perhaps you
need to network a number of Quadras in a desktop publishing
environment where users are continually transferring lOM
Photoshop files, or perhaps you need to upgrade a number of CAD
users that are continually taxing the network. In these cases, the
extra bandwidth and throughput of Ethernet or FDDI can pay for
itself in short order.
If you’re planning future network growth and expansion, don’t
invest heavily in the daisy-chain approach. If you’re planning to
expand, consider instead scenario 3b, which uses an active
Ethernet hub.
Chapter Fourteen Design and Implementation
Scenario 2: Single LocalTalk Network
Using an Active Star Topology
If you plan to stay with LocalTalk but would like to avoid the
hassles associated widi daisy-chaining, then a star topology is the
solution (see figure 14.5). Apple’s LocalTalk devices will not work
with this topology; you’ll need to use the phone-type connectors
(PhoneNET or equivalent). (We'll continue to use the term
LocalTalk for convenience.) You can also use Farallon’s Star-
Connectors. They do not require termination and are a bit less
expensive than the conventional two-port devices. The easiest and
cheapest LocalTalk star is known as a passive star. The passive star
simply joins together the free ends at a common point. This can be
a telephone-style punchdown block or a patch panel.
Figure 14.5
A single Locallalk
neiwoikwiiti a passive
stariopology.
This approach makes sense for a small workgroup. You may be
able to use existing building wiring to create your passive star.
Considering this potential for reduced wiring costs, coupled with
the low cost of the StarConnectors, this network scenario can be
the most affordable of all. The shortcomings of passive stars are the
limits imposed on the number of devices (3 branches with 10
devices each) and the limited extensibility.
To summarize the considerations:
+ Ideal for small workgroups in close proximity
+ Can use existing building wiring
Part Four Network Design, Implementation, and Management
- Restrictive wiring limitations and limited flexibility
- Limited to approximately 30 devices
• Maximum cabling length of 1,000 feet
Figure 14.6
A single locallalk
neiwoik using an aciive
startopnlngy.
Scenario 3a: Single LocalTaik Network
Using an Active Star Topology
The alternative star topology is the active star (see figure 14.6). It
uses a multiport LocalTaik repeater to separately feed each seg-
ment of the star. Active stars are generally more reliable and easier
to maintain than their passive counterparts. Each cable run has a
maximum length of 3,000 feet. For a twelve-port repeater, the
maximum cabling distance is 36,000 feet. Most LocalTaik star
repeaters are equipped with management software that can enable
or disable ports and perform basic line quality testing.
LocalTaik Repeater
While it’s possible to place forty (or even more) devices on an
active star, it’s important to remember than the entire star is a
single AppleTalk network, sharing the LocalTaik bandwidth of
230.4 Kbps. So, unless your network demands are modest, it’s best
to limit the number of nodes to 30 or 40.
The cost of an active LocalTaik star network is more than the
passive star because you must factor in the cost of the repeater.
Chapter Fourteen Design and Implementation
Depending on the number of nodes and the cost of the repeater,
this could add a considerable amount to the per node cost.
To summarize the considerations:
+ Star topology makes it easier to add and move devices
+ Preferred over passive star topologies in most cases
+ Enhanced reliability
- Limited to 30 to 40 devices
- The cost of the repeater increases the cost per node
• Maximum cabling length of 3,000 feet per segment
Scenario 3b: Single Ethernet Network Using
a Star Topology with lOBase-T Cabling
This scenario provides the bandwidth of the thinwire scenario,
with the flexibility of a star wiring topology (see figure 14.7).
Twisted-pair (lOBase-T) Ethernet networks require a repeater hub.
These hubs come in many different sizes and price points. There
are even low-cost mini-hubs that have anywhere between four and
eight ports.
Figure 14.7
A single Eilieinet network
using a star topology and
10Dase-I cabling.
As in the case of the LocalTalk stars, you might be able to use
existing building wiring, although lOBase-T wiring requires four
Part Four Network Design, Implementation, and Management
pairs of wires that meet certain requirements. If you must create
your own wiring, don’t worry, it’s not too difficult — lOBase-T
Ethernet uses a larger version (RJ-45) of the RJ- 1 1 connectors used
by the phone-type LocalTalk connectors. You can buy the bulk
cable, connectors, and proper crimping tools from network supply
houses such as Black Box. (Consider the use of Level 5 wiring just
in case you plan to migrate to a high-speed network such as 100
Mbps Ethernet or CDDI.)
This single hub approach makes a lot of sense when you need the
extra bandwidth of Ethernet and are planning for future growth.
When you require additional hubs, the hubs can be linked together
with an Ethernet backbone. Of course, the cost of hubs add to the
per-node cost, but that’s the price of the additional flexibility.
The length of each twisted-pair segment cannot exceed the 100
meter restriction, so the maximum distance between devices
(assuming the hub is centrally located) is 200 meters.
To summarize the considerations:
+ The star topology makes it easier to add or move devices
+ Uses low-cost wiring
- Additional cost of the hub adds to the overall cost
• Maximum lOBase-T length of about 325 feet
Scenario 4: Single LocalTalk Network
Using a Multiple Star Topology
In an ideal situation, you should only place one device on each
port of a LocalTalk star repeater. This way, you’ll be able to isolate
problem nodes more easily. However, if you place one device per
Chapter Fourteen Design and Implementation
port, you will only be connecting a dozen or so devices per re-
peater. You can extend this somewhat by linking the star repeaters
together (see figure 14.8). For example, by interconnecting three
12-port repeaters, you’ll be able to connect 36 LocalTalk devices
without having multiple devices per port.
LocalTalk Repeater
- (Farallon, Focus)
Figure 14.8
A single locallalk
neiwoikusingamuliiple
siariogology.
Of course, this is an expensive way to connect LocalTalk nodes. In
the example of the three linked repeaters, the per-node cost might
exceed $100. Keep in mind that even though multiple repeaters are
used, the resultant network is still a single logical LocalTalk seg-
ment with a shared bandwidth of 230.4 Kbps. Due to the relatively
high per-node cost and the inherent limitations of LocalTalk, be
leery of this scenario.
The considerations are:
+ Avoids multiple devices per repeater port
- The LocalTalk bandwidth is shared by all devices
- Still limited in practicality to 30 to 40 devices
- Cost per device is relatively high
• Maximum cabling length of 3,000 feet per segment
Part Four Network Design, Implementation, and Management
Figure 14.9
A single Locallalk
neiwoik using a bridged
slariopoingy.
Scenario 5: Single LocalTalk Network
Using a Bridged Star Topology
An alternative to the LocalTalk repeater is Tribe’s LocalTalk bridge
(see Figure 14.9). This device uses packet switching technology to
maximize the limited bandwidth of LocalTalk. If the low-cost and
easy installation of LocalTalk appeals to you, but you’re finding the
LocalTalk bandwidth limiting, then Tribe’s LocalSwitch might
provide a good alternative. It increases the throughput and extends
the number of devices that you can place on your LocalTalk
segment. According to Tribe, you should be able to put a total of
60 devices on the bridge.
The considerations are:
+ The bridge maximizes the limited LocalTalk bandwidth
+ No router administration is required
- Limited to 40 to 60 devices
- Limited extensiblity without an Ethernet connection
• Maximum cabling length of 3,000 feet per segment
Scenario 6: Multiple LocalTalk Networks
Using a Serially Routed Topology
Eventually, as you continue to add LocalTalk devices to the net-
work, you'll reach a point where separate AppleTalk networks are
required (see figure 14.10). This might be due to the sheer number
Chapter Fourteen Design and Implementation
of devices or the need for traffic isolation. The key to this scenario
is the LocalTalk-to-LocalTalk router. There are a number of com-
panies that make such a router. The router could also be a
Macintosh that’s running the Apple Internet Router through
LocalTalk connectors in the modem and printer ports. In this
example, each LocalTalk network is implemented with a repeater
or a Tribe bridge.
Figure 14.10
Mulliplelocallalk
networks serially louled.
The router connects each LocalTalk network in a daisy-chained
manner, but it’s wise to limit the number of connected networks to
a maximum of 4 or 5 networks (3 or 4 routers). Anything more than
this increases the risk of network timeouts and delays. (The router
introduces a bit of administrative overhead due to the configura-
tion of its network numbers and zone names.)
This scenario makes sense as an upgrade path from the single
LocalTalk network that is becoming burdened with traffic. This is
particularly the case when the additional traffic can be isolated by
the addition of the router. The only shortcoming with this ap-
proach is that it’s somewhat limited when it comes to expansion. A
much better approach is to use an Ethernet network as a backbone
(see Sceneuio 7 a below).
To summarize the considerations:
+ The router isolates each LocalTalk segment
• Each LocalTalk network supports 30 to 40 devices (using a
repeater)
Part Four Network Design, Implementation, and Management
• Each LocalTalk network supports 40 to 60 devices (using a
bridge)
- Daisy-chaining networks is practically limited to 3 or 4
routers
- Limited expansion capability
• Maximum cabling length of 3,000 feet per segment
Figure 14.11
Multiple locallalk
neiwoiks uslnparouieil
(not tuidoeill backbone
topology.
Scenario 7a: Multiple LocalTalk Networks
Using a Routed Backbone Topology
When you outgrow your single LocalTalk network, or when you
have to connect LocalTalk devices to Ethernet-equipped devices,
the easiest way is to link the LocalTalk networks to an Ethernet
backbone (see figure 14.1 1). Each LocalTalk network will connect
to the backbone with a LocalTalk-to-Ethernet AppleTalk router.
With this approach, you’ll be able to interconnect hundreds of
LocalTalk devices. It still makes sense to use some kind of star
repeater or bridge in combination with the router. Some vendors
(such as Farallon) offer combination devices that merge a
LocalTalk-Ethernet router with a star repeater.
As in the prior scenario, some router administration is required.
If you plan to connect a number of routers to the backbone, you
may want to consider tire use of a seed router. Seed routers are
used to load non-seed routers with network numbers and zone
Chapter Fourteen Design and Implementation
information. Most routers offer this seeding capability, imple-
mented as a part of the router's configuration. The primary advan-
tage of seeding is administrative. Changes made to the backbone’s
network number range or zone names can be made in a single
router instead of changing all the routers on the backbone.
To summarize the considerations:
+ Ethernet backbone provides high-speed path
+ Extensible to many routers and hundreds of devices
+ Traffic isolation with routers
- Router administration required
- LocalTalk bandwidth — no increase in throughput
• Ethernet (thickwire) backbone length approximately
1,500 feet
Scenario 7b: Multiple LocalTalk Networks
Using a Bridged Backbone Topology
If you anticipate fewer than 254 devices, then another solution is
available from Tribe (see figure 14.12). Their TribeStar bridge has
an Ethernet connection in addition to several LocalTalk ports. The
TribeStar has the advantage of reduced administration because
there are no routers to configure and maintain.
TribeStar
— LocalTalk-Ethernet \ f
Bridge \
Rgure 14.12
Multiple iocallalk
neiwoiksusioga
bridged — not
routed — backbone
topology.
Part Four Network Design, Implementation, and Management
This scenario makes sense for medium-sized LocalTalk networks
where the maximum performance must be obtained for a mini-
mum of administrative overhead. The Ethernet backbone provides
a measure of extensibility, so that future bridges (assuming the 254
node limit is not reached) and routers can be added as needed.
The considerations are:
+ The Ethernet backbone provides a high-speed path
+ Extensible to many bridges and hundreds of devices
+ No router administration required
+ Maximizes LocalTalk bandwidth
• Each bridge can support up to 64 devices
- More than 254 nodes requires an AppleTalk router
• Ethernet (thickwire) backbone length of about 1,500 feet
Scenario 8: Ethernet Backbone with Multiple LocalTalk
Networks and Direct AppleTalk Nodes
Building on Scenario 7a, you can always add AppleTalk nodes
directly to the Ethernet backbone (see figure 14.13). This approach
can be used to connect Ethernet-equipped Macs and LaserWriters
to the network. It is also widely used to connect other computers,
such as UNIX workstations, DEC VAX, and PCs to the network of
Macs.
Chapter Fourteen Design and Implementation
If AppleTalk is to be the transport of choice, then the router can be
an AppleTalk-only unit. If other non-AppleTalk nodes are added
to the Ethernet, then the router must be a multiprotocol router
capable of routing the necessaiy' protocols between the LocalTalk
and Ethernet devices.
Keep in mind that the LocalTalk and Ethernet repeaters have
different cabling length requirements. LocalTalk runs may ap-
proach 3,000 feet, while lOBase-T limits are about one-tenth that
distance (100 meters, or about 325 feet).
The considerations are:
+ Ethernet backbone provides a high-speed path
+ Extensible to many routers and hundreds of devices
+ Direct Ethernet connections with thickwire or thinwire hubs
• Ethernet (thickwire) backbone length about 1,500 feet
• lOBase-2 length about 600 feet; lOBase-T length about 325
feet
Scenario 9; Multiple Ethernet Networks Using
a Routed Backbone and Star Topology
As the cost of Ethernet connections continue to fail, and as Apple
continues to offer built-in Ethernet in more and more Macintosh
models, the trend is clearly away from LocalTalk connections. For
this reason, solutions like Scenarios 7a and 8 are beginning to
become less popular. The direction that most network designers
are taking is to use Ethernet to all desktop devices, and then to use
an Ethernet segment as a connecting backbone (see Figure 14.14).
The key component is an Ethernet-to-Ethernet AppleTalk router.
These devices are becoming more prevalent. Cayman, for example.
Part Four Network Design, Implementation, and Management
Figure 14.14
Mulliplellliernei
netwoiksusinga
backbone and Slat
toonlogy.
has just released such a router; other companies are likely to follow
suit. The router could also be a Macintosh with two Ethernet cards
running the Apple Internet Router.
10Base-2(Thin) or
10Base>T Repeater
Ethernet-Ethernel
AppleTalk Router
This scenario represents the most flexible and high-performance
option for all but the most demanding applications. This scenario
also fits in nicely with structured wiring plans, which makes it easy
to respond to growth and redistribution of network resources; it
also makes it easy to add routers to the network in response to
excessive traffic. This approach lends itself to intelligent hubs and
concentrators that work in conjunction with, or incorporate,
AppleTalk routers. As an example, Cabletron has just announced
their intention to incorporate die Cisco 4000 series router into their
hubs. Standalone high-performance multiprotocol routers are also
likely candidates, as they provide the necessary performance with
a wide range of protocol support. These routers are supplied by
several vendors; Cisco and Wellfleet are two popular choices.
To summarize the considerations:
+ Ethernet backbone provides a high-speed path
+ Extensible to many routers and hundreds of devices
+ Full Ethernet bandwidth between nodes
• Ethernet (thickwire) backbone length approximately
1,500 feet
• lOBase-2 length about 600 feet; lOBase-T length about
325 feet
Chapter Fourteen Design and Implementation
Scenario 10: Multiple Ethernet Networks
with a FDDI Backbone
As the number of routers increase on an Ethernet backbone, its
ability to handle the inter-netvvork traffic plus the inter-router
traffic can become strained. For these very large networks, replac-
ing the Ethernet backbone with FDDI is becoming more prevalent
(see figure 14.15). For now, the main role of FDDI is likely to be its
use as a high-speed backbone medium.
Figure 14.15
UullipleEiliefnei
networks will) a FOOl
backliDne.
The latest generation of high-speed routers provide, in addition to
Ethernet and Token Ring ports, FDDI ports that can provide the
connection to the FDDI LAN. Since FDDI LANs can extend over
great distances, this scenario is indicated for very large networks
with many attached subnetworks.
To summarize the considerations:
+ FDDI backbone provides a high-speed path
+ Extensible to many routers and hundreds of devices
+ Full Ethernet bandwidth between nodes
• FDDI supports 1,000 nodes, each up to 2km (over 1 mile)
apart, for a total aggregate distance of 100km (over 60 miles)
Scenario 11: Ethernet and FDDI WAN Topology
The last wiring scenario involves the combination of Scenarios 9
and 10 by adding a high-speed WAN connection (see figure 14.16).
Part Four Network Design, Implementation, and Management
These WAN connections are commonly made with the multi-
protocol routers mentioned earlier. As mentioned before, one of
AppleTalk’s past shortcomings was the regular transmission of
routing table updates. These updates tended to be burdensome to
AppleTalk WAN links. Today, there are many options to solve this
problem.
Figure 14.16
Etheinei and 1001 WAN
lopologv.
Ethernet Backbone
First, Apple has added the AURP routing protocol, which only
sends routing updates when necessary. AURP support is included
with Apple’s Internet Router, and currently being added to many
third-party routers. Another solution for WAN AppleTalk support is
through the encapsulation of AppleTalk within another protocol,
such as TCP/IP. The process of IP encapsulation of AppleTalk has
been recently defined and standardized. Lastly, many high-end
router developers offer specialized routing protocols that can be
applied to the routing of AppleTalk over the WAN.
It is expected that Apple will take additional steps in the near
future to enhance AppleTalk’s viability over wide-area networks.
These include the support of upcoming Point-to-Point Protocol
(PPP), which should enhance remote access and routed connec-
tions, and the adoption of other popular routing protocols such as
Open Shortest Path First (OSPF), which is popular in large IP
internets.
The considerations can be summarized as follows:
+ FDDI backbone provides a high-speed path
Chapter Fourteen Design and Implementation
+ Extensible to many routers and hundreds of devices
+ Full Ethernet bandwidth between nodes
• FDDI supports 1,000 nodes, each up to 2km (over 1 mile)
apart, for a total aggregate distance of 100km (over 60 miles)
Scenario 12: Structured Wiring Example
This last example illustrates a wiring scenario that uses a technique
sometimes referred to as a structured wiring implementation (see
figure 14.17). In this example, a star topology is used to wire all
offices and cubicles vdth several runs of lOBase-T (or even Level 5
compliant) twisted-pair v\iring. All wires converge at a master
patch panel within a wiring closet. From this patch panel, connec-
tions are made to the appropriate wiring devices based on require-
ments of the connected device.
Wiring Closet Office Cubicles
19" Rack
LocalTalk Hub &
Router or Bridge
Terminal Server
10Base-T Hub
Figure 14.17
Anetampleola
situciuied wiling
implemeniaiion.
LocalTalk devices are interconnected to a LocalTalk hub and
router; Ethernet devices are connected to an Ethernet hub. While
Part Four Network Design, Implementation, and Management
it’s not shown in figure 14.16, the lOBase-T hub could also be
connected to an Ethernet-to-Ethernet router (as shown in Scenario
9). Serially connected devices are patched to a terminal server.
What's important about this scheme is that all devices use identical
wiring. Changes are confined to the wiring closet. If a LocalTalk
Mac Plus is replaced with an Ethernet Quadra, the only required
change is to move a patch cord from the LocalTalk hub to the
Ethernet hub.
Some integrated hubs merge the different hubs (such as terminal
servers, lOBase-T, and even LocalTalk) into a single, unified
chassis. These devices make the wiring process even easier and less
cluttered by eliminating the cross-connects that are now part of
the hub’s backplane.
Conclusion
The development and implementation of an AppleTalk network
is not a one-time activity. It is an ongoing process; the network
continues to evolve as users are added and technology changes. In
this chapter, a number of network scenarios were outlined and
discussed. Each successive scenario provided additional complex-
ity— and additional capabilities — over the previous examples.
Collectively, these examples delineate a full range of options for
the reader; they also illustrate a clear and progressive growth path
from the simplest of LocalTalk networks to the most complex
combinations of Ethernet, Token Ring, and FDDl cabling.
Management and
Troubleshooting
ow many copies of MacDraw do the users own?
H| Macs have enough memory to run System
H iH Why can’t the users on the second floor print
^8 to the LaserWriter? What happens to our network
performance at 10 AM every day? Chapter 15
discusses common questions like these — and
provides answers to them.
Configuration Management
Keeping track of the software and hardware status of every
Macintosh on a network is not an easy task. How much memory
does Mary’s Mac have? Is everyone using the same version of
PageMaker? Many potential problems can be avoided if every user
on the network has the same version of important applications and
system software, and they all use the same collection of fonts and
font technology (TrueType or PostScript). Several products are
now on the market that allow you to examine and even install
software updates on another Macintosh, using the network to
transfer the information. These products are often marketed under
Part Four Network Design, Implemenation, and Management
the misnomer “network management" tools, perhaps because the
person holding the title Network Manager is often responsible for
configuration management of every node on the network. These
products, while quite useful, do not address the task of actually
managing the network itself; they manage the Macs connected
to it.
Figure 15.1
lelwofked'floile
management
How Do They Work?
A node management product (see figure 15.1) requires that you
install a small program, usually called a responder or agent, on
every Macintosh you will manage. The network manager’s
Macintosh then runs an application which communicates with the
responder using a private format. In many products, the responder
can also handle the task of copying updated versions of system
or application software to the correct location on the remote
machine’s hard disk. This feature alone can justify the cost of the
product, because it will save an enormous amount of time and
effort every time you have to upgrade software on every node on
your network.
Management Node
Managed Node
Be sure to pick a product that can upgrade the responders, or you might still
find yourself roaming the halls with an installer disk a few times a year.
Chapter Fifteen Management and Troubleshooting
Server and Account Management
Another labor-intensive task is controlling who can use which file
servers and printers, who can use the network’s dial-out or dial-in
services, and reporting when or how often each one has used each
service. These resources always have specialized requirements, so
each has its own dedicated management tools. Whenever possible,
select products which can be managed across the network, instead
of those that must be managed locally.
Network Management
At the present time, AppleTalk network management is undergo-
ing significant changes. In the past, AppleTalk management
worked with any number of third-party router configuration
programs to inspect and modify AppleTalk network numbers and
zone names. Several years ago, Apple began development of a
management protocol called the Apple Management Protocol, or
AMP. During the development process, Apple made the decision to
abandon AMP and adopt the industiy standard Simple Network
Management Protocol, or SNMP. When the promise of SNMP for
AppleTalk is fully realized, managers will have a consistent and
universal interface for network management.
Simple Network Management Protocol (SNMP)
SNMP is a popular tool for network management in large TCP/IP
networks. SNMP provides a means for a management station to
query a remote device for configuration information, and in some
cases, to load new configuration information. For example, SNMP
can be used to examine the Address Mapping Table of an Apple-
Talk router, or to find out how many DDP packets (or bytes) were
sent and received by an Applet alk node.
Part Four Network Design, Implemenation, and Management
The information that SNMP maintains is defined in a data struc-
ture definition called the Management Information Base (MIB).
The complete MIB for AppleTalk is defined in an Internet Engi-
neering Task Force Request For Comment document, RFC- 1243.
An SNMP management station communicates with an SNMP
agent running on a managed entity (a network node such as a
router or gateway). In order for data to be transferred between the
devices, it must be described within the MIB. Most MIB elements
are currently read-only, so most of the management can be more
accurately described as monitoring. As the security features of
SNMP improve, more elements in the MIB will allow read-write
access from management stations, so active management of
AppleTalk network devices using SNMP will become a reality.
For example, the AARP MIB defines access to the information kept
in a node’s AppleTalk address mapping table. Each entry in the
table matches one hardware address to one protocol address, and
identifies the port on which the information is used. Since this is
the only AMT information in the MIB, it is the only AMT informa-
tion a management station can request of an SNMP-compliant
AppleTalk node.
Similar MIBs are defined for the other lower-layer AppleTalk
protocols, such as DDP, RTMP, and ZIP. By examining the MIB
definitions, you can see precisely what information is available for
each protocol, and what configuration details (if any) can be set
remotely.
Because SNMP is a Format layer protocol, there is no reason why it
can’t be used over AppleTalk Instead of TCP/IP. Apple now in-
cludes SNMP protocol support in the Macintosh network drivers
for LocalTalk, EtherTalk, and TokenTalk. This means that
AppleTalk nodes can now maintain statistics on how busy they are
and report that information on request to an SNMP management
Chapter Fifteen Management and Troubleshooting
station. Figures 15.2 and 15.3 depict where the differences lie
between these two implementations of SNMP.
Management Node Managed Node
Management Node Managed Node
Hgure 1 5.2
SNMP with ICP/IP
iranspon.
Figure 15.3
SNMP will) Applelalli
iranspoiL
By sending an alert to the management station, SNMP also pro-
vides a means for nodes to notify the network management station
of events, such as error conditions. A limited number of these
alerts are defined in the current AppleTalk RFC, but the number is
expected to grow.
Currently, there is one SNMP management program, called
WatchTower from InterCon, that runs on the Macintosh. Because
it was developed before a standard was defined for using SNMP
over the AppleTalk transport, it uses TCP/IP as the message
transport protocol. As SNMP via AppleTalk becomes more popular,
SNMP management products using AppleTalk will certainly
become available.
Part Four Network Design, Implemenation, and Management
Troubleshooting
Diagnosing problems on an AppleTalk network can be a very
complicated task. Fortunately, there are software tools that can be
a great help in investigating and solving network problems, but no
one tool can do the entire job. If you’re serious about diagnosing
AppleTalk, you’ll also need a copy of Inside AppleTalk, Second
Edition, which contains Apple’s official, definitive description of
the AppleTalk protocol family.
How the Chooser Really Works
The Chooser is the user’s window onto the network, so it often
provides the first indicadon of impending network problems. For
example, a user might report that a machine can’t find a parUcular
printer or some other resource in the Chooser; perhaps the
Chooser doesn’t show any zones at all. To solve AppleTalk problem
reports, you must know the topology of your network (how the
routers connect everything) and you need to know where the
affected devices are located.
The Network Census — Who Is Out There?
Apple Computer’s lnter»PolI is an easy-to-use network tool (see
figure 15.4). This application uses the NBP protocols to obtain
both the NVE (the object, type, and zone names) and the logical
AppleTalk addresses of the AppleTalk devices found on the net-
work. An accurate NVE and address list is a great help whenever
you’re looking into a network problem.
However, whenever you use Inter»Poll, keep in mind that NBP is
an “unreliable” means of locating every node on the network. The
list it produces may not include every node on the network be-
cause some devices simply do not respond to NBP Lookups, while
others may not respond if they are busy when the lookup message
Chapter Fifteen Management and Troubleshooting
arrives. When Inter»Poll is left to run long enough, and the
“Unnamed Devices” option has been selected and added to the
search list, you will usually see at least one entry (the unnamed
socket number 4) for every active node on the network.
-Select Zone(s);-
Chicago, IL
•ir
EthorTalk
TokonTalk
Total Zones: 5
O nil Named Deuices
(§) Deuices Matching:
Match
a
•Search
✓ Net
Name
Type
<Unnome(f>
Searching for AM Devices
in Zone: New York, NY
•Search Time:-
[EUnin
□ Continuous
Sec
-Select Sorttng:-
Net
3
(i) Rscending
O Descending
With
® Partial Match
[ leaf
rn^
ar \m
[ Can< ei )
K )1
Figure 15.4
Apple's imei'Polluiiy.
If the printer which could not be seen in the Chooser appears in
lnter»PoU’s list, you can then test how long it takes to send a packet
to the printer and back. Within lnter*Poll, this is accomplished by
double-clicking on the line containing the name of the printer. A
dialog box appears that enables you to send a request for the
printer status; Inter»Poll displays the results of the test (see figure
15.5). More importantly, you can send a series of echo packets (the
AppleTalk Echo Protocol) to the printer, and Inter*Poll will report
how long it took to return each one. Because NBP has a two-
second timeout, if packets are lost or take longer Uian two seconds
to return, the Chooser isn’t likely to display the printer in its device
list. Notice that the display also tells you how many hops tlie reply
message took to return to Inter* Poll. If the number of hops varies
widely, the device may be too many router-hops away.
Part Four Network Design, Implemenation, and Management
Figure 15.5
lnier«Pollcantesi
how long ii lakes to sent
a packet loihegrinlei
and back.
Figure 15.6
Faiallon'sNeiAtlas
creates a logical network
magolApplelalk
networks.
DeviC8:
N*t:7097 Nod# : 158
‘s L«s#rVrU#r - L*s#rVrit#i* - Ovol Offic# Zone
Pockets:
20
Using;
Intervol:
2.5
— OEchoPkts
Secs (i) Printer Status Packets
Timeout;
1.5
s*cf O System Info Packets
I( )l
( Done ]
Rovd: 8 Lost: 0
Pockets Sent: un: 12 toui; 8
Current
Average
Minimum
Maximum
Hops Av/eg
7
7.00
7
7
DeUg (secs)
0.15
0.16
0.13
0.20
[ Clear ]
Slotus:
status ; idW
Network Mapping Programs
Another very helpful diagnostic tool are programs that draw a map
of your network by interrogating devices actually on the network.
LANsurveyor from Neon Software, and NetAdas from Farallon (see
figure 15.6), are two Macintosh network mapping products. These
programs can only create an accurate map if the information they
receive from the network’s routers is correct. If a mapping program
is unable to draw a map of your network, or if it produces some-
thing very different than what you expected, you should determine
the reason. This technique is often a useful diagnostic eiid.
Chapter Fifteen Management and Troubleshooting
Router Troubles
As vve discussed in earlier chapters, routers play a ver>' important
role in the operation of the Chooser, and in network performance
in general. As the number of routers on a network increases, so
does the challenge in tracing and diagnosing network problems.
Remember that all routers must agree on the network number
range and zones list configured on each router port. Every time a
new router is added to an active network, there is a chance that it
will conflict with the existing routers on the network. If this hap-
pens, a product which allows you to examine the active configura-
tion of any router on the network, such as Neon Software’s
RouterCheck (see figure 15.8), can help you locate the source of the
configuration conflict. RouterCheck can also notify you (either on
screen or by text pager) whenever a trusted router experiences a
network change, such as adding or dropping a routing table entry
or zone.
! Pacfcet #287 I
FIo9S 0x80 902.3
Statu*. 0x00
Packet Length 64
Tli*«*taftp 0.39:27.424
Etrwfrn«t Heoder
Ckij 1 1 nation
Sourc*
LLC Length
8g2.2J<tfl<Kr
Drt t
Source SftP:
Control
Protocol
09:00 07 ffffifr
08:00 90 01:73:47
40
OXCKI
Oxoa
0x03
0x080007800b
Phase 2 Broodcost
AppleTalk
LflftO PtW* Hooder - Datooroa DpI Ivtm f*rotoe«l
Unused
Hop Count
Datograe Length
OOP Check sue
Dest Heteork
Sote^e heteork.
Oest Mode
Source Hode:
Oest Socket
Sois-ce Socket.
OOP Type:
soo
*0000
32
0x0000
0
2000
253 0 255
too
I RTrtP Socket
I RThP Socket
I RThP Response cr Data
ftPPleTalk Brdcasl
BTHP Pocket - Botitifwi Toiiie Wointenonce rrokocot
Router's Met
2000
10 L*r>gth
8
Router s liode 10
0xc7
too
RTWP Tuple «t
RorKje St.TTt
2000
Range Flog;
*100
Extended
Dist<r»c*
0
Ror^ie End
2000
Uerslon
0x82
RTI1P Tuple «2
flci'ige Star*.
2000
Range Flog
*100
Extended
Distance
0
Range End
2000
Oersion
0x82
BTRP Tuple «3
lkrlxoi> l*ijfber
2017
Range Flog
*000
honex tended
Distance
0
00 00 00 00 00 00
Frane Check Sequence 0xd8o2ef4‘;1
:sm
Figure 15.7
Afouieidieckusinii
NeonSoflwaie's
flouieiChetk.
Part Four Network Design, Implemenation, and Management
Figure 15.8
itoPeek capture Ola
leniryRIMP packet.
Notetltetedundant
eptries.Iliisoccuts
with tlie Shiva FastPaili
louters and does not
cause any ptoblems.
Listen and Learn
The most informative network diagnostic tools, sometimes called
netivork analyzers or packet sniffers, enable you to examine the
packets, or frames, that actually appear on the network. Until
recently, all network analyzers were very expensive, single-purpose
hardware instruments, beyond the budget of the typical manager
of a small- to medium-sized network.
Fortunately, there are now very good Macintosh-based network
analyzers. These programs convert your ordinary Macintosh into a
LocalTalk, Ethernet, or Token Ring network analyzer for a fraction
of the cost of the dedicated instruments. NetMinder LocalTalk
and NetMinder Ethernet are products of Neon Software; the AG
Group’s network analyzers are LocalPeek, EtherPeek, and
Chapter Fifteen Management and Troubleshooting
TokenPeek. As their names imply, these products are Cabling
layer- or DataLink layer-specific, and (unlike most Macintosh
programs) they aren’t expected to work with every possible combi-
nation of Macs and networking cards. These programs are topically
designed to work with selected high-performance network cards
and adapters. Confirm that the hardware requirements of the
product match the configuration of the system you intend to use.
A network analyzer is used to capture and examine the packets that
are being sent over the netw'ork. This is the network equivalent of
eavesdropping on someone else’s conversation, and we all know
how revealing that can be. And yes, these analyzers can be used to
read unencrypted data and perhaps learn passwords and other
confidential information. Many applications use simple encoding
schemes to avoid sending dear-text messages across network
links, specifically to make reading packet contents a bit more
difficult. This is one area where Apple’s Open Collaboration
Environment will help. AOCE will provide a new encrypted trans-
port protocol called Apple Secure Data Stream Protocol (ASDSP).
A common problem when adding a new router is that the new
router refuses to start, reporting a conflicting network range or
zones list. Assuming that you’re sure you have configured the new
router correctly, how can you determine the source of this conflict-
ing information?
If the network range is in conflict, you can simply capture the
Routing Table Maintenance Packets (RTMP), as shown above in
figure 15.8, and examine them for conflicting information. Every
router broadcasts an RTMP packet every 10 seconds, so it
shouldn’t take long to find the culprit’s physical hardware and
logical protocol addresses. You can then use information from
lnter»Poll, or use the hardware address, to identify the errant
device. Simply rebooting the router may clear the problem, or it
may be necessary to reconfigure it.
Part Four Network Design, Implemenation, and Management
Diagnosing Chooser Problems
Consider a small EtherTalk network that has several Macs and a
LaserWriter. Let’s look at what actually happens when you open
the Chooser to select a LaserWriter.
You may recall from Chapter 8, "Macintosh Transport: AppleTalk,"
that the Chooser loads its AppleTalk Zones window by using the
Zone Information Protocol (ZIP) to request a list of the reachable
zones from a local AppleTalk router. Wlien you click on tlie
LaserWriter icon in the Chooser, it obtains the list of active printers
in the selected zone by sending NBP Lookup messages to the
selected zone. As NBP Response packets containing the names of
currently active printers arrive, the printers’ names will be added
to the list of names displayed on the right side of the Chooser. Let’s
examine these Zone Information and Name Binding messages in
detail to see exactly how they work.
Figure 15.9 shows a trace (captured by NetMinder Ethernet) of all
the messages in and out of a Macintosh as the Chooser was opened
and the LaserWriter icon was subsequently selected. This trace is
quite small because a filter was used to ignore all the other traffic
that was present on the network at the same time.
Figure 15.9
Chooser iraflic as viewed
bvNelMiodet.
Chapter Fifteen Management and Troubleshooting
NetMinder has formatted the headers of the captured packets so
we can easily see the address of the node sending the request
(source), the address the message was sent to (destination), and
the type of packet see figures 15.10 through 15.13).
Packet 0 (figure 15.10) is a “GetZoneList” (6ZL) request, sent from
node 1800.228 (the Macintosh) to node 1800.194 (an EtherTalk
router).
Figure 15.10
PackeiO.
lust 16 milliseconds later, the router sent Packet 1 (figure 15.1 1), a
“ZoneInformationProtocolReply” (ZIPR), which contains the list of
zones on this network.
In this case, the entire list could not fit into one reply message (the
reply contained 43 zones), so the node sent a second request
asking for the next portion of the zone list, starting at number 44;
this was Packet 2 (figure 15.12).
Part Four Network Design, Implemenation, and Management
Figure 15.11
Packet).
Pacitel I
File Edit Search Options Filter Ullndoms nnoiysis
NetMjnder6 Ethernet
Size
CED E 1
10 3666
~64
1800.194 1600.228
RTP-OZL 1
2
2 183
64
1800.194 1800.228
flTP-OZL
3 16
198
1800.228 1800.194
flTP-ZIPB
4 1616
73
1800.194 1800.228
NBP-Brftq
5 0
64
PppleJ5488d5 Rpple-d470f7
BRRP-Rpl
6 0
90
1800.228 1800.44
H8P-Rep
7 1466
73
1800.220 1800.228
HBP-BrPq
8 0
90
1800.228 1800 44
HBP-Rep
5
Total Packets
266 Status Stopped
Total Fittered
0 Save File
Total Errors
0 Buffer Use 0%
1 Buffered PMkats
9
C
in Size; 602
T <ndi>- 3603 (1/29/93 S 36:11 Pti
Errors : Hone
Ethornot Header
OesUrMUon Opp I e.jd470f?
Source C i seo_02797d
Length $0248
flTalk Phase 2 Header
DSPP $oo
ssnp $oa
Control $03
Protocol RTolk
Long OOP Header
Length
Checksum
Oest Het
Sre Het
Dest Mode
Sre Node
Dest Socket
Sre Socket
Tgpe
576 <Hops-0)
$6d0
1800
1800
228
194
245
6
3 <ffTP>
RTP/ZIP Header
CiKl/CntI $90 (TBesp EOM >
BiMp/Seq $0
TID $319
Lost Flog 0
Nua Zones 43
D
Preu
Nent
Figure 15.12
PackeiZ
Packet 2
File Edit Search Options Filter tilindoiiis Analysis
0
1
3666
16^^
64
602
1800. 194
1800.228
1800.228
1800. 194
RTP-OZL
RTP-ZIPR
3
16
198
1800.228
1800 194
RTP-ZIPfl
4
1616
73
1800. 194
1800.228
HBP-erRq
5
0
64
RppleJ5488dS
f^le^TOf?
RRRP-Rpl
6
0
90
1800.228
1800.44
HBP-Rep
7
1466
73
1800.220
1800.228
NBP-BrRq
8
0
90
1600.228
1800.44
HBP-Rep
NetMlndBr€' Ethernet
mmm \asmm mumi \mmii
1 12] Size: 64
T <i»s)« 3866 <1/29/93 5:36:11 PH
Errors: Hone
Ethernet Header
Destination Ciscoj02797d
Source Rpple^47Qf7
Length $00 Id
nialk Phase 2 Header
OSflP $oa
SSRP $QQ
Control $03
Protocol fiTolk
Long OOP Header
Total Packets
266
Status
Stopped
Total Filtered
0
Save File
Total Errors
0
Buffer Use
0%
Buffered Packets
9
Length
Checksvn
Dest Het
Sre Net
Dest Hode
_J Sre Node
•n- Dest Socket
Sre Socket
Type
21 (Kops-O)
$0
I8X
1800
194
228
6
245
3 (ATP)
RTP/ZIP Header
Ced/CntI $40 (TPeq >
B«op/Seq $ 1
TIO $3ta
Function GetZoneList
Start Index 44
The second reply from the router, Packet 3 (figure 15.13), con-
tained 13 additional zones. At this point, the Chooser has enough
information to display a complete zone list.
Chapter Fifteen Management and Troubleshooting
^ File Edit Search Options Filter lUindoun nnolysls
NclMinder<) Ethernet
mmm rnnm^
[ED
Size
nm
E
|[ irUr. j'.l-cfk 1
0
36M
64
1800 194
1800.228
flTP-OZL
1>
1
15
602
1800 228
1800. 194
ftTP-ZIW
18^^
64
1800 19^^^
ftTP-OZL
4
1616
73
1600 194
1800 228
MBF-8r8q
5
0
64
nppl«J5488(f5
Pppl*_d470f7
WlFP-npl
6
0
90
1800.228
1800 44
HBP-Rep
7
1466
73
1800.220
1800 228
MBP-Brfiq
8
0
90
1800 228
1800 44
HBP-Reo
1 Total Packeia
266 Status Stopped
1 Total Filtered
0 Save File
1 Total Errors
0 Buffer Use OS?
1 Buffered Pockets
9
T <iis>- 3883 (1/29/93 5:36 1
Errors: None
Ethernet
Header
OeatirKitian
Ppple.jd470f7
Source
Ctscoj02797d
Length
S00t>4
fllalk Phase 2 Head
OSRP
$oa
SSf»>
too
Control
m
Protocol
RTalk
Long IIDP
* Header
Lertgih
172 <Hops-0>
ChecKrun
$tt>42
D«st Net
1800
Src Net
1800
Dost Node
228
Src Node
194
Oest Socket
245
Src Socket
6
Type
3 <RTP)
RTP/2IP Header
Ced/CntI
B»ap/S«q
TIO
Last FIo 9
Nue Zones
V30 <TR«sp EOtl >
SO
S3 In
llill'
Prew
Figure 15.13
Packei3.
A little more than a second later — 1.616 seconds, to be exact —
when the user clicked on the LaserWriter icon, an NBP-BrRq (NBP
Request) was sent to the router. This was Packet 4 (figure 15.14).
Figure 15.14
Packet 4.
Part Four Network Design, Implemenation, and Management
If the router is directly connected to a network in the target zone, it
will broadcast an NBP Lookup message onto the network; all active
LaserWriters will then send an NBP reply containing their names
back to the Macintosh. This trace doesn’t show these broadcasts
from the router, because for this example we have configured
NetMinder Ethernet to capture only messages to and from the
Macintosh running NetMinder.
If the routing table in the router contains entries for distant net-
works using the same zone name, it will fomard the NBP Request
to those other networks, where a local router will convert the
request to a broadcast message and send it out on any local
network which is in the target zone.
A LaserWriter on the local network (1800) has heard the router’s
broadcast, and will now send a reply to the Mac. In order to send
an NBP reply directly to the Mac, rather than back through the
router, the LaserWriter must learn the correct physical hardware
(that is, Ethernet) address to use to send a packet to the Mac’s
logical protocol address (AppleTalk address 1800.228).
Recall that the AppleTalk Address Resolution Protocol (AARP) is
used to associate a node’s Ethernet address to its AppleTalk
address. The information necessary to do this is kept in an Address
Mapping Table (AMT) in every EtherTalk node. The LaserWriter
checked its AMT and found that no entry exists for node 1800.228,
so the LaserWriter broadcasts an AARP probe/request containing
the Mac’s logical AppleTalk address.
The Macintosh responded to the AARP request with the AARP reply
(AARP-Rpl) shown in Packet 5 (figure 15.15). The LaserWriter
added the information to its AMT and can now send EtherTalk
NBP replies directly to the Macintosh. Packet 6 (figure 15.16) is the
first of these.
Chapter Fifteen Management and Troubleshooting
Figure 15.15
Packets.
m FHe Edtl Search Options Filter llUndoiiis Analysis
Netvilndei€» Etheinet
I PacIcBt 6 I
:emi
j~
(cmcaa Iwinw iMSir
n
t
hum
I riarii 1
81 za
In’ki-' Ifia’.kfi 1
1
rr^
E
0
3666
~64
1800 194
1800.228
RTP-OZL
1
16
602
1800 228
1800. 194
RTP-2IPR
2
183
64
1800 194
1800.228
RTP-OZL
3
16
108
1800 228
1800 194
RTP-ZIPR
4
1616
73
1800 194
1800.228
teP-BrRq
5
0
64
RpplaJ5488d3
Rppla^70f7
HIH
■DBDHHHH
BlBHIHiH
HBDIHH
■
7
1466
73
1800 220
1300 228
MBP-erRq
8
0
00
1800 228
1800 44
rCP-Atp
o
1 Total Packets
266 Status Stopped
1 Total Fttlered
0 Save Fite
1 Total Errors
0 Buffer Use OSS
1 Buffered Packets
9
J
L
RTolk Phoao 2 Header
OSAP too
S6AP iaa
Corilrol 103
Protocol RTolk
Long DDP Header
Langth
Diacksub
Oast Mat
Src Mat
Oast Moda
Src Moda
Oast Sockat
Src Socket
Tgpa
64 (Hops«0>
SO
1800
1800
228
44
254
2
2 ««P>
HBP Header
Control S3 (LkUp-naoly)
Tupla Count
rCF ID
It
126
HBP Tuplea
1800
la 44
kat 128
e II
Rppla fttO <I8/60U> (L-18)
« LosarUritar <L>1I>
m MV Rppla RtO <L-I2>
Rgure 15.16
Packets.
The NBP request and reply packets will continue as long as the
Chooser remains open on the Macintosh. Under System 7, the
Part Four Network Design, Implemenation, and Management
longer the Chooser remains open, the less frequent the NBP
request will be sent, eventually dropping to just one every thirty
seconds.
With a packet-level view of the network, and perhaps a trace of a
normal example to compare against, you can often determine if a
problem is caused by the Macintosh, a router that isn’t doing what
its supposed to, or a device which isn’t responding as it should. If
the printer doesn’t send the NBP reply, then the printer is obvi-
ously the problem. If a router doesn’t send a zone list when the
Mac requests one, then the router is the problem. In cases where
the problem conversation crosses networks or zones, it may be
necessary to trace the messages at each end (or even on each
network along the way) in order to get a complete picture of what
is happening.
Performance
Network performance is a complex equation consisting of many
variables. Some of these variables, such as bandwidth, are easy to
quantify. Other variables, such as traffic, are more ambiguous.
Networks are dynamic, unpredictable, complex systems. Optimiz-
ing their performance requires a basic understanding of the
underlying theory, access to traffic monitoring tools, and the
ability to interpret — and act upon — test results.
The Real Meaning of Bandwidth
Everyone wants his computer to operate as quickly as possible.
As described in previous chapters, Ethernet is not faster than
LocalTalk at the physical level. All signals travel at nearly the speed
of light. But not all network interfaces send and receive signals at
the same speed. How can both of these statements be true?
Chapter Fifteen Management and Troubleshooting
Ethernet hardware can create and detect a single electrical state
change (corresponding to a binary digit) in l/40th the time of
LocalTalk. So an Ethernet can carry 40 times more information
than LocalTalk in a given period of time. But a single Macintosh on
Ethernet typically communicates only three or five times faster
than on LocalTalk. Why this discrepancy? It’s because the
Macintosh cannot send and receive data at Ethernet’s maximum
rate. The most important performance benefit of Ethernet over
LocalTalk is that, because each bit can be sent (and received) so
quickly, the cable is much more likely to be idle at the moment
your computer needs to use it, even with many more nodes on the
network.
Traffic Jams
How can you tell if network traffic levels are affecting user perfor-
mance? A relatively easy way is to use a neUvork analyzer such as
EtherPeek or NetMinder to trace the signals on the network. After
collecting a sample of all the traffic on your netv\'ork, these pro-
grams will tell you exactly how busy your network really was at that
time.
Benchmarks
It’s a good idea to do this occasionally when the network is operating
normally, so as to obtain benchmarks of your normal network state.
Another technique is to install specialized devices that listen to the
traffic on the network and count how many packets and bytes were
sent from each node on the network, and to what destination.
Dayna sells specialized pods that can be connected to a segment in
order to monitor the traffic levels. The pod can be queried by a
program called Network Vital Signs, which uses the data reported
by the pods to provide information about the monitored network.
Part Four Network Design, Implemenation, and Management
Any network which shows a high utilization is likely to force nodes
to wait to send data often enough to affect performance. In many
cases, network traffic may be high only a few times throughout the
day. By learning as much as possible about these traffic peaks, you
may be able to distribute or isolate these demands and avoid
paying for expensive network upgrades.
Analysis of the trace may show that the majority of the traffic is
caused by just a few nodes that talk to each other constantly. In
this case, moving these top talkers to their own network may bring
performance back to acceptable levels for everyone.
Another means to monitor network utilization by individual nodes
is to have each device keep track of how much data they transfer
and to report the information, upon request, to a central station
that compiles information from many nodes to create an over-all
picture of the network’s utilization. SNMP management stations
can poll SNMP managed nodes regularly throughout the day and
compile this information into a time-line analysis of network
utilization.
Traffic Cops
On LocalTalk, adding a router between the section with the top
talkers and the rest of the network is the most common way to
prevent the traffic between certain nodes from delaying traffic
among the other nodes. As an alternative, traffic can be isolated by
adding a LocalTalk switch or bridge, such as Tribe’s LocalSwitch,
that dynamically connects multiple LocalTalk segments only when
the source and destination of a particular packet requires it. Tribe
recently announced a new product called TribeStar. It has eight
bridged LocalTalk ports and one Ethernet port.
On Ethernet and Token Ring networks, link-layer bridges can also
be used to isolate traffic within limited portions of the netw'ork.
Chapter 5, “Common Network Components,” described how an
Chapter Fifteen Management and Troubleshooting
Ethernet bridge examines the source and destination of tlie physi-
cal hardware addresses on every packet it hears on each port and
maintains a table of known addresses that have been heard on
each port. Wlien the bridge receives a packet on one port, it will
resend it on the other port unless the destination address is known
to be on the same port as the sender. In this case, the bridge does
not transmit the packet on the other port, and the traffic level on
the rest of the Etliernet is reduced.
Traffic levels across the entire Ethernet can be significantly re-
duced by moving top-talking pairs onto the same side of strategi-
cally placed bridges. Afterwards, the node’s connectivity to the rest
of the Ethernet will be unchanged. Wlienever a packet’s destina-
tion address is on the other side of the bridge, it will be sent. In
effect, the bridge is an imdsible gate that allows packets to pass
through only on a need-to-go basis.
It is sometimes possible to achieve utilization of 200% to 300% of
Ethernet’s total bandwidth through the intelligent installation of
Ethernet bridges (see figure 15.17).
Hgure 15.17
Ethernet bridoes.
Consider the network diagram in figure 15.17. The three pairs of
nodes. A, B, and C, each consume up to 40% of the Ethernet
bandwidth communicating with each other, and up to 10% com-
municating with all the other nodes on the network. Without
bridges, this Ethernet would experience peak demands of 150% of
the available bandwidth. But, by adding two bridges in tlie appro-
priate places, no single section will ever see a demand higher
that 60%.
Part Four Network Design, Implemenation, and Management
In Token Ring environments, Source Routing Bridges (SRBs)
provide a similar opportunity to isolate high volume conversations
onto their own ring. Unlike Ethernet bridges, SRBs are not entirely
invisible at the link level. Their presence requires some additional
overhead because ring-to-ring routing information must be added
to every packet. But SRBs are invisible at the AppleTalk protocol
level, so they can provide some traffic isolation which is invisible to
the user.
Another possible source of performance problems are the routers.
As inter-network traffic grows, the routers will become unable to
keep up with the demand. Most routers keep simple counters of
how many packets are sent and received, and how many were
ignored or discarded due to various errors.
Figure 15.18 shows a statistic screen from an Apple Internet
Router, running on a Macintosh Ilci. The counters indicate that
this router is very lightly loaded, but the Recent Network Error Rate
is extremely high; the Network Reliability is only 63.4%. Notice that
the error counter Local Net Setup Conflicts is already 22 on the
Ethernet port. This indicates that there is a router with a conflicting
configuration on the Ethernet. Eliminating the conflict will reduce
the Error Rate and Network Reliability will begin to improve.
Figure 15.18
Houier Errors.
Port Statistics for Router Mo
Packet) Routed:
38 Network Reliability:
63.45?
Recent Activitg Rate:
1 1 1 j i 1 1 1 1 1 1 1 1 1 1 Recent Network Error Rate;
hiiilHiiiiin
Idle
Busy
Low High
Statistics last reset at: Sat, Jan 30, 1 993
10; 12 PM
Statistic
Total
ISI
m
LocalTeIc
EtherT^lk
Packets In
19
14
5
Packets Out
19
14
5
Name Requests In
19
19
0
Name Lookups Out
19
14
5
Data Link Errors
0
0
0
Packet Buffer Overflow
0
0
0
Unknown Network
0
0
0
Hop Count Exceeded
0
0
0
Routi ng Table Overflow
0
0
0
Local Net Setup Conflicts
22
0
22
Remote Net Range Conflicts
0
0
0
Router Version Mismatch
0
0
0
±1 11
* 2 ]
Chapter Fifteen Management and Troubleshooting
One way to determine the source of these conflicts is to capture all
the RTMP and ZIP packets on the segment for a few minutes. By
examining the RTMPs sent by each router, the one sending a
conflicting Network Range can be identified. By examining the
Zone Information Protocol (ZIP) messages, a router sending a
different zones list can be identified.
On an AppleTalk network, security is primarily the responsibility
of the individual devices. AppleTalk File Sharing and AppleTalk
Remote Access use the "Users & Groups” file to manage access to a
node. AppleShare servers have a more advanced version of the
same scheme. At the other end of the spectrum, Apple Laser-
Writers (PAP servers) don’t restrict network access at all.
Some network devices also support additional forms of security,
restricting what network services can be seen by a particular node.
AppleTalk Remote Access can be configured to grant access to
“just this node” or the entire network. Some routers can be config-
ured to hide or show any device, or entire class of devices (for
example “Joe’s AppleShare” or “All LaserPrinters”) based on the
identity of the node seeking the device. Routers from Compatible
Systems do this by forwarding NBP messages to and from your
node only if you’ve properly identified yourself to the router and if
the router’s manager has granted permission for your node to
communicate with the device (or device type) of interest.
A similar scheme is utilized by several non-Macintosh AppleTalk
Remote Access products. One is AsyncServeR, a software ARA
server for VAX/VMS, that supports individualized network access
profiles for each dial-in connection account; anodier is the Shiva
LanRover family of dedicated ARA servers.
Part Four Network Design, Implemenation, and Management
A few high-end routers, such as those from Cisco, can be config-
ured to perform true zone hiding. In this case, the router hides
zones from all die other routers on the network. Over wide-area
connections, this feature can be used to limit Chicago’s view of
Denver’s network to a single zone, rather than the thirty-five zones
that actually exist in Denver. Instituting a limited view of remote
networks can be useful in controlling which devices at one site can
communicate with devices in other sites. It can also be used in very
large networks to simply reduce the number of zones in the
Chooser to a more manageable number.
These schemes have one significant limitation. When a router
provides the security, only connections that come through the
router can be controlled. Access attempts originating within the
local network remain uncontrolled and unrestricted.
A security feature of AOCE (Apple Open Collaboradon Environ-
ment) will provide an AppleTalk network-wide access control
mechanism. AOCE allows a network administrator to grant or deny
access to a network device (or type) for an individual user, or group
of users.
Conclusion
If you manage an AppleTalk network, there is no substitute for
understanding AppleTalk, but there are quite a few tools that,
when used together, can help reveal many of AppleTalk’s secrets to
you. This chapter has described only a small pordon of them.
The Future of
Macintosh
Networking
ne goal of this book was to give the reader a “gut-
level" understanding of the theories, compo-
nents, and implementation details involved with
Macintosh networking. Hopefully, as new
services, formats, transports, and cabling sys-
tems come along, you'll be better equipped to
lentify, classify, and evaluate their suitability for
your networking environment.
Actually, it seems as if each week brings new networking tech-
nologies. While it’s hard to predict the future, some trends in
Macintosh networking are likely to continue.
One such trend is wireless networking. You should expect an
explosion in this technology over the next few years. This growth
should occur throughout all the various wireless technologies —
cellular, radio waves, and infrared. These technologies should
primarily affect the implementation of local-area networks and
remote users.
0
Part Four Network Design, Implementation, and Management
Another area to watch in the near future is the continued enhance-
ment of the AppleTalk protocol stack. It’s been widely discussed by
Apple, and in the press, that the AppleTalk Remote Access Protocol
will be enhanced to work over the Point-to-Point Protocol (PPP).
This will have the effect of lowering the cost and enhancing the
performance of dial-up access. Other AppleTalk changes are likely
to affect the routing aspects of the protocol. The newly released
AURP routing protocol is likely to undergo continued updates;
there’s even talk that Apple will adopt more sophisticated routing
algorithms such as the OSPF (Open Shortest Path First) protocol.
It’s also likely that Apple will continue the enhancement of
AppleTalk by making some of the traditional AppleTalk services,
such as AFP and PAP, protocol-independent. Perhaps not too far in
the distant future, you will be able to access these services over the
TCP/IP protocol stack instead of the AppleTalk stack.
Even the Macintosh-compatible cabling systems are likely to
evolve over the years. One obvious change is that Apple’s inclusion
of Etliernet is likely to extend further down the product line into
the midrange, and perhaps even into the low-end systems as well.
As Ethernet finds its way into the mainstream product lines, expect
high-bandwidth replacements for Ethernet to appear. At the
moment, it’s somewhat unclear which of the new high-speed
technologies — either FDDI/CDDl or 100 Mbps Ethernet — is likely
to become the dominant LAN cabling system. In either case, the
cost is likely to plunge as economies of scale are applied.
All in all, when it comes to networking, Apple’s in an extremely
strong position for the future. At every layer, Apple has continued
to innovate. At the Service layer, Apple has pioneered dynamic
service acquisition and continues to provide world-class services.
The two most recent examples of this are Publish and Subscribe
and AOCE. At the Format layer, Apple has acknowledged the
importance of format transportability with the Clipboard, XTND,
Chapter Sixteen The Future of Macintosh Networking
and EasyOpen. Apple’s attention to the Transport layer is well
known. From AppleTalk’s dynamic addressing to MacTCP’s
integration, Apple consistendy provides transport excellence. Even
at the Cabling layer, Apple continues to provide innovation, with
LocalTalk, Ethernet, and with various v\ireless technologies.
Glossary of
Networking
Terminology
O n the following section, you might notice the
icons beside many of the definitions. The icons
are representations of the four layers of the
NetPlCT model:
Services layer
Formats layer
Protocol layer
Cabling layer
Each icon indicates the layer of the NetPICT model to which the
defined term belongs. Using these icons, you can easily determine
where each term fits in the networking universe. If a term can be
specifically placed within the seven-layer OSl Reference model,
then its OSI layer is indicated within the definition.
Live Wired
lOBase-2 An Ethernet implementation that uses a thin
coaxial cable. lOBase-2, like other Ethernet implementa-
tions, has a bandwidth of 10 Mbps. The maximum cable segment
length is 185 meters.
lOBase-5 The original Ethernet medium, lOBase-5 uses a
heavy shielded coaxial cable. The cable has an approxi-
mate diameter of 3/8", It has a bandwidth of 10 Mbps. The maxi-
mum segment length is 500 meters.
lOBase-T An implementation of the Ethernet standard
that runs over two pairs of unshielded twisted-pair
wiring. Like the other Ethernet implementations, lOBase-T has a
bandwidth of 10Mbps. The maximum segment length from the
hub to the device is 100 meters.
802.3 An IEEE standard that defines the CSMA/CD
(Carrier Sense Multiple Access/Collision Detection)
method of network access. CSMA/CD is used by Ethernet net-
works.
802.5 An IEEE standard that defines the Token Ring
network access metliod.
-A-
AARP See AppleTalk Address Resolution Protocol.
AAUI See Apple Attachment Unit Interface.
access privileges Controls placed upon network services
that limit and control user access.
Glossary of Networking Terminology
active star A network with a multiport repeater at die
center. Each device connects to the repeater. Active stars
do not perform network addressing — network packets seen on one
branch of the star are seen on all branches.
address A sequence of bits used to identify devices on a network.
Each network device must have a unique address. Addresses fall in
two categories: physical hardware addresses (Ethernet), and logical
protocol addresses (AppleTalk).
Address Mapping Table (AMT) A table that
associates physical hardware addresses with
corresponding logical protocol addresses. In the case of AppleTalk,
the AMT is updated and maintained by the AppleTalk Address
Resolution Protocol (AARP).
address resolution The association of physictil
hardware addresses (Ethernet) with logical
protocol addresses (AppleTalk). See also Address Mapping Table.
Address Resolution Protocol (ARP) A TCP/IP
protocol used to map logical IP addresses onto
physical hardware addresses (Ethernet). Similar to Apple’s AARP.
ADSP See AppleTalk Data Stream Protocol.
AEP See AppleTalk Echo Protocol.
AFP See AppleTalk Filing Protocol.
amplitude The magnitude of an electrical signal. Measured by
subtracting the minimum voltage from the maximum voltage of an
electrical signal.
AMT See Address Mapping Table.
Live Wired
analog A method of data transmission where (unlike digital
transmissions) the data is continually modulated over an infinite
voltage range.
Apple Attachment Unit Interface (AAUI) Apple’s physi-
cal Ethernet interface. AAUI uses a special connector
and uses an external transceiver, either thickwire, thinwire, or
twisted pair, to connect to the network.
AppleShare file server A Macintosh product from Apple
that runs the AppleTalk Filing Protocol (AFP).
AppleTalk Apple’s networking software that provides reliable
delivery of data between clients and servers. There are implemen-
tations of AppleTalk on many different computers.
I AppleTalk address A three-part number that uniquely
identifies a particular network, node, and socket on an
AppleTalk network.
AppleTalk Address Resolution Protocol (AARP)
Apple’s protocol that maps the logical
AppleTalk node addresses to the physical hardware addresses.
I AppleTalk Data Stream Protocol (ADSP) A connection-
oriented Session layer protocol that provides a reliable,
bi-directional stream of data between two sockets in an AppleTalk
network.
AppleTalk Echo Protocol (AEP) An AppleTalk Transport
layer protocol that enables a node to send a special packet
to another node and to receive an echoed response in return. Used
to determine round-trip delivery times and reachability.
Glossary' of Networking Terminology
AppleTalk Filing Protocol (AFP) The AppleTalk Presen-
tation layer protocol that defines shared file access.
Platform-independent, AFP is the basis for Apple’s AppleShare
product.
AppleTalk for VMS An implementation of the AppleTalk network
protocol suite that runs on a DEC VAX under the VMS operating
system. It essentially turns a VAX into an AppleTalk node. Applica-
tion services, such as the VAXshare file server, are seen as
AppleTalk sockets.
I AppleTalk Session Protocol (ASP) An AppleTalk Session
layer protocol that provides for the creation of a network
session. It also keeps the communications in the proper sequence.
AppleTalk Transaction Protocol (ATP) An AppleTalk
Transport layer protocol that manages the give and take
of a network transaction.
AppleTalk-LAT connection tool ACTB (Communications
Tool Box) tool that is used in conjunction with the
AppleTalk-LAT Gateway. It passes AppleTalk protocols to the
gateway, where they are then converted into DEC’S LAT protocol.
Used to access LAT terminal services over LocalTalk or AppleTalk
Remote Access sessions.
AppleTalk-LAT Gateway A software gateway that runs on
a Macintosh and is used to translate AppleTalk protocols
into LAT protocols. The AppleTalk-LAT Gateway is part of DEC’S
PATHWORKS for Macintosh product.
AppleTalk/DECnet Transport Gateway A software
gateway that is included with AppleTalk for VMS. It
provides Mac users with access to DECnet-based applications,
such as E-mail and DECwindows.
Live Wired
Application layer The topmost layer of the ISO's OSI
Reference Model; it corresponds to the Service layer of
the NetPICT model. It defines the protocols and connections for
applications. See OSI Reference Model.
ARCNET Originally developed by Datapoint Corpora-
tion, ARCNET (which stands for Attached Resource
Computer NETwork) is a popular LAN in the IBM PC world.
ARCNET uses a token-passing scheme and runs on coaxial or
twisted-pair wiring. Several vendors make ARCNET cards for the
Macintosh.
ARP See Address Resolution Protocol.
ASCII ASCII (which stands for American Standard Code
for Information Interchange) is a commonly used coding
scheme that uses 8 bits of data to encode alphanumeric characters
and special control characters. In many ways, ASCII text files are a
common denominator among many computers and programs.
ASP See AppleTalk Session Protocol.
Asynchronous communication (Also known as async
communication.) A technique of data transmission that
sends one character at a time without waiting for an acknowledg-
ment.
ATP See AppleTalk Transaction Protocol.
attenuation The loss or diminution of an electrical
signal that occurs during transmission.
AWG AWG (which stands for American Wire Gauge) is
the US standard for specifying the diameter of a wire
conductor.
Glossary' of Networking Terminology'
backbone network A network topology where devices
connect to a single cable. Thickwire Ethernet networks
are commonly used as backbone networks.
bandwidth The total message-carry'ing capacity of a
medium. Bandwidth is typically measured in bits per
second. It is not an indication of speed.
baseband A kind of network transmission that uses the
entire bandwidth of a medium to transmit a signal.
Baseband communications are commonly used by most LAN
cabling systems, such as LocalTalk, Ethernet, and Token Ring.
bits per second (bps) A unit that measures the message-carrying
capacity of a medium. A kilobit per second (Kbps) is one thousand
bits per second; a megabit per second (Mbps) is one million bits
per second.
bps See bits per second.
bridge A network device used to connect two netw'orks
at the Data Link layer. Bridges are essentially unaware of
the logical protocol address, although some bridges can block
protocols by filtering their type codes.
broadband A kind of network transmission that splits
the bandwidth of a medium to support multiple chan-
nels of communication. This technique is used by cable television.
broadcast transmission A network transmission that is
sent to ail network devices.
Live Wired
brouter A network device that routes the
routable protocols and bridges the non-
routable protocols. It essentially merges the functionality of
bridges and routers.
bus A common network segment. Network devices
connect to the same segment. See backbone network.
bus topology A network scheme that uses a single cable
to connect devices. Unlike ring topologies, the cable
does not connect to itself. See backbone network.
-c-
Carrier Sense Multiple Access with Collision Avoidance
(CSMA/CA) A cable access technique used by LocalTalk.
Devices listen for the presence of a carrier before transmitting.
Nodes try to avoid collisions by backing off for a random period
of time whenever a message is not successfully transmitted.
LocalTalk doesn’t have the detection and recovery circuitry found
in CSMA/CD networks.
Carrier Sense Multiple Access with Collision Detection
(CSMA/CD) A cable access technique used by Ethernet
that allows devices to gain access to a transmission medium by
listening for the presence of a carrier. If no carrier is detected, the
data is transmitted. Each node is capable of detecting collisions
and retransmitting as required.
cheapernet A slang term describing thinwire (lOBase-2)
Ethernet.
Chooser The Macintosh desk accessory used to select
AppleTalk (and other) services.
Glossary of Networking Terminology
client A client is a process or entity that employs the
services of other processes or entities knowoi as servers.
For example, a Macintosh uses AFP client software to access AFP
servers.
client/server A term given to the interaction of software
processes that function in a cooperative manner. Clients
make requests of servers.
coax See coaxial cable.
coaxial cable A cable that contains two conductors, one
inside the other. Coaxial cables, sometimes called coax,
are used in thickwire and thinwire Ethernet and in IBM terminal
connections.
collision The condition that results when two network
devices transmit at nearly the same lime. The transmis-
sions collide, rendering the message unusable.
Control Panel A Macintosh utility that enables you to
change various aspects of the Macintosh. The Netw'ork
control panel is used to select from different network data links.
This is where the cabling choice (such as EtherTalk, LocalTalk,
TokenTalk, or FDDITalk) is made.
CSMA/CA See Carrier Sense Multiple Access with Collision
Avoidance.
CSMA/CD See Carrier Sense Multiple Access with Collision
Detection.
Live Wired
-D-
Data Link layer The second layer from the bottom of the
OSI Reference Model. The Data Link layer corresponds
to the top half of the Cabling layer of the NetPICT model. The Data
Link layer defines the protocols that manage the creation of
network frames, such as LLAP (LocalTalk Link Access Protocol).
See OSI Reference Model.
I datagram A packet of data, unique to a specific protocol.
Datagrams are placed within network frames for delivery
over the network. With AppleTalk, the Datagram Delivery Protocol
(DDP) is used to encapsulate higher-level AppleTalk protocols
within an addressable unit.
Datagram Delivery Protocol (DDP) The Network layer
protocol that provides the ultimate addressing of
AppleTalk datagrams (network, node, socket) over an AppleTalk
network.
DECnet A suite of networking protocols developed by
Digital Equipment Corporation for use on VAX/VMS,
PDP-11, PCs, and other computers. Versions of DECnet are avail-
able for the Mac from Digital and Thursby Software Systems.
DECnet tunnel A technique used by AppleTalk for VMS
and DECnet, which encapsulates AppleTalk datagrams
within DECnet packets for delivery over a DECnet link.
default zone The AppleTalk zone that a device belongs to
by default when it’s placed on an extended (Phase 2)
network. The default zone of a Mac can be changed by double-
clicking on the current network icon in the Network Control Panel.
Glossary of Networking Terminology
dynamic node addressing A technique Apple uses
whereby AppleTalk nodes automatically pick unique
network addresses. This is in direct contrast to other protocols,
such as DECnet, that maintain a listing of node addresses for each
node.
-E-
EBCDIC EBCDIC (which stands for Extended Binary
Coded Decimal Interchange Code) is a coding system that
uses 8 bits of data to represent alphanumeric characters and
control sequences. Used by IBM mainframes.
ELAP See EtherTalk Link Access Protocol.
emulation See terminal emulation.
entity type The part of an NVE (Network Visible Entity)
that identifies the generic class to which the entity be-
longs. Examples include "AFPServer" or “LaserWriter." The entity
type is assigned by Apple Computer, who maintains a registry.
Ethernet A LAN cabling system originally developed by
Xerox, Intel, and Digital. Ethernet has a bandwidth of
10Mbps and uses the CSMA/CD access method. Ethernet supports
many different networking protocols, including AppleTalk.
EtherTalk Apple’s implementation of AppleTalk proto-
cols on Ethernet. EtherTalk places the DDP datagrams
into Ethernet frames.
EtherTalk Line Access Protocol (ELAP) The specific data
link protocol used by EtherTalk.
Live Wired
FDDI FDDI (which stands for Fiber Distributed Data
Interface) is a 100 Mbps LAN technology' that uses a
token-passing access method. FDDI uses dual fiber-optic rings.
There are several FDDI cards available for the Macintosh. A
variation of FDDI that uses copper wiring, called CDDI, is currently
under development.
file server Software that provides network users with
shared, controlled access to files. In the Macintosh
environment, the standard for file service is the AFP file server.
File Transfer Protocol (Fl'P) A Service layer protocol
common in the TCP/IP world. It is used to copy files
between network devices.
frame A data link structure for conveying information
over a transmission medium.
frequency The number of times in a unit of time (usually a second)
that a signal cycles between minimum and maximum voltage.
FTP See File Transfer Protocol.
gateway A device (or software process) that translates entire
protocol stacks.
Glossaty of Networking Terminology
half-router A pair of routers connected with a communi-
cation link. This tandem acts as a single routing device.
AppleTalk half-routers require unique network numbers on each
side.
hardware address The unique physical address deter-
mined at the Physical and Data Link layers. For example,
each Ethernet card has a unique hardware address that's stored
within the card.
hop A unit of distance that measures the passage of a
datagram through a router. The distance between net-
works is often measured by tlie number of hops.
hop count The number of routers that a datagram travels
through on the way to its destination. AppleTalk permits a
maximum of 15 hops, although hop count adjustment can be used
to provide routing flexibility.
International Standards Organization (ISO) An international
standards-making body responsible for numerous standards,
including film speed and the OSI Reference Model.
internet The term applied to multiple AppleTalk networks that are
connected by AppleTalk routers. Also called an internetwork.
internet router (IR) A device that connects AppleTalk
networks by using network numbering to pass, or deny
passage of a datagram.
Live Wired
Internet Protocol (IP) The Network layer protocol of the
TCP/IP protocol suite. Analogous to Apple’s DDP.
Internetwork Packet Exchange (IPX) A Network layer
protocol used by Novell NetWare that provides address-
ing, routing, and switching packets.
IP See Internet Protocol.
IPX See Internetwork Packet Exchange.
IR See internet router.
ISDN ISDN (which stands for Integrated Services Digital
Network) is a digital communications standard that
integrates voice and data. ISDN services are just beginning to
become common in the US. It offers the promise of affordable
high-speed WAN communications.
ISO See International Standards Organization.
-K-L-
Kbps Kilobits per second. See bits per second.
LAN See Local Area Network.
LAP See Link Access Protocol.
LLAP See LocalTalk Link Access Protocol.
Glossary of Networking Terminology
Local Area Network (LAN) A network in one area, such as
a building or a group of buildings.
Local Area Transport (LAT) DEC’S proprietary, licensed Ethernet
protocol, used to connect terminal devices to host computers.
Macs can speak the LAT protocol directly with the LAT Communi-
cations Tool. At the present time, LAT’s use is limited to terminal
emulators and terminal front-ends.
LocalTalk Apple’s low-cost LAN network that runs over
twisted-pair wiring. LocalTalk has a bandwidth of
230,400 bps (230.4 Kbps).
LocalTalk Link Access Protocol (LLAP) The Data Link
level protocol that manages the delivery of data on a
LocalTalk network.
Management Information Base (MIB) A database used
by SNMP for maintaining the status and control for a
netw'ork device. Each network device has its own specific MIB.
Data within the MIB is used by an SNMP agent as part of a network
management application. AppleTalk MIBs have been recently
defined by Apple and are currently being implemented by most
AppleTalk networking vendors.
Mbps Megabits per second. See bits per second.
MIB See Management Information Base.
live Wired
modem A device that converts digital data from a computer into
analog data, which can then be transmitted over a telephone line.
This process is called modulation. It also performs the opposite
process, demodulation, to convert incoming analog signals into
digital data that the computer can understand.
multicasting Multicasting is similar to broadcasting, but
it provides a protocol-specific method of identifying
netw'ork devices. Each protocol, such as AppleTalk, has its own
multicast address.
-N-
Name Binding Protocol (NBP) The AppleTalk Transport
level protocol that maps the NVE onto the corresponding
AppleTalk address. NBP provides AppleTalk with a way to dynami-
cally find named services on the network.
NBP See Name Binding Protocol.
Network File System (NFS) A Presentation layer protocol
developed by Sun Microsystems to provide TCP/IP
networks with file sendees. NFS is similar in scope to Apple’s AFP.
Network interface controller (NIC) A networking card,
such as an Ethernet, Token Ring or FDDI, for a
computer.
Network layer The layer of the OSI Reference Model that
controls network addressing. The Network layer corre-
sponds to the bottom third of the Protocol layer of the NetPICT
model. With AppleTalk, the Network layer is defined by DDP. IP is
the Network layer for TCP/ IP networks.
Glossary of Networking Terminology
network number A 16-bit number used to identify the
AppleTalk network to which a node is assigned. Nodes
choose their network number from an AppleTalk router; when no
router is present they choose a number from a pre-defined range
known as the startup range.
network number range A range of network numbers that
have been established within the routers for use on Phase
2 extended network segments. Non-extended networks, such as
LocalTalk, are restricted to single network numbers.
Network- Visible Entity (NVE) An AppleTalk concept that
names devices by their name, generic type, and zone.
Examples include “My Mac:Macintosh Quadra 950@BlueZone”
and “Joe’s Printer:LaserWriter@NYC Zone.”
NFS See Network File System.
node An addressable entity on a network, such as a Mac
or a LaserWriter.
node identifier (node ID) An 8-bit number that uniquely
identifies each node on a single AppleTalk network
number.
non-extended network An AppleTalk network that supports
addressing of up to 254 nodes and supports only one zone. A
LocalTalk network is an example of a non-extended network.
NVE See Network- Visible Entity.
Live Wired
-O-P-
OSl Reference Model A model put forward by the International
Standards Organization (ISO) that provides a standard point of
reference for networking protocols. It uses seven layers to break
down die network process into independent processes. (OSI
stands for Open Systems Interconnection).
packet An organized sequence of binary data that includes data
and control structures.
PAP See Printer Access Protocol.
passive star A network topology where every branch is
connected to a common point. Unlike active stars,
passive stars have no repeater at the center to actively retransmit
signals.
peer A network device that is treated as the communicative equal
to another device on the network. Networks that let devices
communicate with each other as equals are sometimes referred to
as peer-to-peer networks.
Physical layer The level at which the actual delivery
medium is realized. For example, the three different
kinds of Ethernet and two different kinds of Token Ring cable are
defined in this layer. The Physical layer of the OSI reference model
corresponds to the bottom half of the Cabling layer of the NetPICT
model.
Glossary of Networking Terminology
Point-to-Point Protocol (PPP) A protocol that supplants
the Serial Line Internet Protocol (SLIP). It provides remote
access to IP networks via asynchronous and synchronous links. It
is similar to Apple’s Remote Access Protocol. There is some discus-
sion that PPP might be adapted to provide remote AppleTalk
protocol access.
port The connection between a router and a network.
Routers can have multiple ports.
PPP See Point-to-Point Protocol.
Presentation layer The layer of the OSl Reference Model
that establishes data formats and requisite conversions.
The Presentation layer corresponds to the Format layer of the
NetPICT model.
print spooler A service that queues print jobs and
manages the submission of these jobs to the printer. This
relieves the client workstation from this task.
Printer Access Protocol (PAP) The AppleTalk protocol
that controls the communication between clients (Macs)
and print servers (LaserWriters).
protocol A set of rules for information exchange over a
communication medium. The set of protocols used by a
particular networking protocol, such as AppleTalk, is called a
family or suite of protocols.
protocol stack (or suite) An implementation of a specific
networking protocol consisting of multiple individual
protocols.
Liue Wired
punchdown block A wiring device used in telephone
and network installations for interconnecting many
wires. Most phone- type punchdown blocks have fifty rows of four
contacts each. The contacts accept the wires, which are “punched
down,” to make electrical contact. A tool, called a punchdown tool,
is used to push the wire onto the contact.
-R-
repeater A network device that repeats the signals on a
network. Repeaters operate at the Physical layer of the
OSI Reference Model. Repeaters amplify weak signals from one
segment and repeat them on another segment.
ring topology A network organization in which all the
nodes are connected in a ring. Data passes around tlie
ring from node to node. Each node retransmits the messages to the
next node in the ring.
RIP See Routing Information Protocol.
router A network device diat connects networks by
maintaining logical protocol information for each
network.
Routing Information Protocol (RIP) A protocol used to
update routing tables on TCP/IP networks. Similar to
Apple's RTMP.
routing table A table, maintained by an AppleTalk router,
that maps the AppleTalk internet by specifying the path
and distance (in hops) between itself and the networks. Routers
use these tables to determine whether or not a datagram should be
forwarded.
Glossary of Networking Terminology
Routing Table Maintenance Protocol (RTMP) An
AppleTalk protocol used by the routers to establish and
maintain the routing tables used by the AppleTalk routers on the
network. RTMP packets are sent at regular intervals by the routers.
RTMP was recently supplanted by AURP, which only sends routing
information when network changes make it nece.ssary.
RTMP See Routing Table Maintenance Protocol.
-s-
SDLC See Synchronous Data Link Control.
seed router An AppleTalk router that contains the net-
work numbers and zone information that is used by non-seed
AppleTalk routers on the network.
segment A length of network cable. A segment can be
connected to the port of a repeater, bridge, router, or gateway.
Sequenced Packet Exchange (SPX) The Transport layer
protocol used by Novell NetWare. Provides a connection-
oriented, guaranteed delivery link between Novell NetWare nodes.
server A network device or process that provides a
service to networked clients. In the AppleTalk environ-
ment, examples of servers include AppleShare file servers and
LaserWriter print servers.
session A logical connection between two network
devices. In an AppleTalk network, the AppleTalk Session
Protocol (ASP) is used to establish, maintain, and discontinue
sessions.
Live Wired
Session layer The layer of the OSI Reference Model that
establishes a logical connection between netw'ork devices.
The Session layer corresponds to the upper third of the Protocol
layer of the NetPICT model.
Serial Line Internet Protocol (SLIP) A TCP/IP-based
protocol used to run IP over serial lines such as dial-up
phone lines. Similar to Apple’s Remote Access Protocol.
shielded cable A cable that is surrounded by a grounded
metallic material. It minimizes disruption of the signal
by external electrical noise and prevents the cable from emitting
unwanted electrical signals.
Simple Network Management Protocol (SNMP) A
management protocol used to maintain and query
network entities. SNMP uses agents on managed nodes to main-
tain a database known as a MIB (Management Information Base).
The data stored within the MIB can be transmitted to the manage-
ment software upon request.
Single Mail Transfer Protocol (SMTP) An electronic mail
service common on TCP/IP networks.
SLIP See Serial Line Internet Protocol.
SMDS See Switched Multimegabit Data Service.
SMTP See Single Mail Transfer Protocol.
SNA See Systems NetworkArchitecture.
SNMP See Simple Network Management Protocol.
socket An addressable entity within an AppleTalk node. Sockets
can be thought of as network processes.
Glossary of Networking Terminology
socket number An 8-bit number that uniquely identifies a socket.
With AppleTalk, there are 256 sockets. 0 and 255 aren’t used,
leaving 254 available for use. Numbers 1-127 are reserved by Apple;
128-254 are available for use by applications.
I split-horizon routing A technique for maintaining routing
tables that was introduced with AppleTalk Phase 2.
Routing information is only forwarded to those routers that can
use the information.
SPX See Sequenced Packet Exchange.
SQL See Structured Query Language.
star topology A centralized network with a hub, concen-
trator, or repeater at the center of the network.
Structured Query Language (SQL) A language, standardized by
ANSI, used to manipulate relational databases. Apple’s Data
Access Language (DAL) uses a variation of SQL to access relational
data through DAL servers.
Switched Multimegabit Data Service (SMDS) A relatively
new high-speed WAN networking technology offered by
telephone companies and service providers.
Synchronous Data Link Control (SDLC) A Data Link
Layer protocol used by IBM’s SNA networks.
Systems Network Architecture (SNA) A suite of commu-
nications protocols developed by IBM. Analogous to the
AppleTalk protocol suite.
Live Wired
-T-
TCP/IP TCP/IP (which stands for Transmission Control
Protocol/Intemet Protocol) is a set of networking proto-
cols commonly found on, but not limited to, UNIX computers.
TELNET A Service layer protocol common to TCP/IP
networks that provides terminal services.
terminal emulation A program, usually on a personal
computer, that masquerades as a computer terminal.
The most common terminals that are emulated are the VT series
established by DEC and the 3270 series from IBM.
terminator A resistive device attached to the ends of a
cable to minimize unwanted signal reflections from the
cable segment.
thickwire A type of Ethernet cabling, also known as
lOBase-5, that uses a thick (approximately 3/8") coaxial
cable. Primarily used as a backbone to which thinwire or twisted-
pair hubs are connected.
thinwire A type of Ethernet cabling, also knowoi as
lOBase-2, that uses coaxial cable and BNC connectors.
TLAP See TokenTalk Link Access Protocol.
Token Ring A cabling system, common in tlie IBM
world, that connects network devices in a ring topology.
It uses the method of token passing to enable nodes to access the
network.
TokenTalk Apple’s product that puts the AppleTalk
protocols onto a Token Ring nehvork.
Glossary of Networking Terminology^
TokenTalk Link Access Protocol (TLAP) The data link
access protocol used in a TokenTalk network. TLAP puts
AppleTalk protocols onto a Token Ring network.
topology The physical connective structure of a network. Com-
mon topologies include bus, ring, and star.
transceiver An interface between a node and the
network. Transceivers transmit and receive network
messages.
Transport layer The layer of the OSl reference model that
ensures that the message is correctly transmitted. If one
portion of the communication transmission is lost or garbled, it’s
the job of the Transport layer to retransmit the necessaiy'^ portion.
The Transport layer of the OSI model corresponds to the middle
third of the Protocol layer in the NetPlCT model.
tunneling A process that enables one protocol’s datagram
to be encapsulated within another protocol’s datagram.
twisted-pair Wiring that consists of two insulated
copper multi-stranded conductors twisted around each
other to reduce electrical interference. Twisted-pair wiring can be
used by phone-type LocalTalk connectors, Ethernet, ARCNET, and
other cabling systems.
VAX A line of computers manufactured by Digital Equipment
Corporation (DEC). VAX stands for Virtual Address extension.
VAXshare Part of DEC’S PATHWORKS for Macintosh
software that enables a VAX to act as an AFP file server
and PAP print spooler.
Live Wired
Vines Banyan’s network operating system. “Vines" is an
acronym for Virtual NEtwork Software.
VMS An operating system used by Digital Equipment Corporation
for its line of VAX computers. VMS stands for Virtual Memory
System.
volume A disk that appears on the Mac desktop. A
volume could be a hard disk, floppy disk, or a network file
server disk.
-w-z-
Wide Area Network (WAN) A network that spans distances beyond
the range served by LANs. WAN distances are usually measured in
miles instead of feet.
ZIP See Zone Information Protocol.
ZIT See Zone Information Table.
zone A way of logically grouping AppleTalk nodes. Zones
are established by AppleTalk routers.
Zone Information Protocol (ZIP) The AppleTalk Session
layer protocol that maintains the mapping of AppleTalk
network numbers to zone names. ZIP is used by the Chooser to
obtain a list of zone names.
Glossary of Netw'orking Terminology
Zone Information Table (ZIT) A listing of zones main-
tained by AppleTalk routers that relates zone names to
the ports of a router.
zone list A listing of AppleTalk zones. Usually seen in tlie
Chooser.
zone name A name assigned to an AppleTalk network
zone. Can be 32 characters long and is case-insensitive.
NetPICT
Encyclopedia
A
his appendix contains a sampling of key NetPICT
diagrams that have been used throughout the
book. These diagrams are also included on the
disk that comes with this book.
AppleShare Server
AppleShare Server
AFP Server (Ethernet)
An AFP server on an Elliernel nelwork. snclr as a Macinlosh
running AppleShare or a VAX running VAXshare.
AFP Server (FDDI)
An AFP server on a FDDI nelwork. An eiample ol such a device
is a Macintosh running AppleShare wilh a FDDI card and the
FDDIFalkpfolocols.
Live Wired
AppleShare Server AppleShare Server
AFP Server (LocalTalk)
An AFP server on a locallalk network, sucit as a Macinlosli
running AppleShare.
LaserWriter Server
Apple LaserWriter (Ethernet)
Same laser printers come equippetl with Ethernet interfaces.
Eiamples inclutle Apple's LaserWriter lip and LaserWriter
Pro 630.
AFP Server (Token Ring)
An AFP server on a loken Ring network. An example wonid he
a Macintosh rnnning AppleShare with a Token Ring card and
the lokenlalk protocols.
LaserWriter Server
Apple LaserWriter (LocalTaik)
Some Apple laser printers come eouipped with locallalk
interfaces. Examples include Apple's oripinal LaserWriter and
the LaserWriter IINI.
Appendix A NetPICT Encyclopedia
Apple Macintosh
Apple Macintosh (Ethernet)
Some Aople Maciniosli commiiers tome equipped wiili Eltieinei
inieifaces as siandaid equipmeni Isuch as ilie Ooadra 9501:
add-oo llheioei cards cao be purchased lor Macs wiihoui buili-
iQlihernei
Apple Macintosh
Apple Macintosh (FDDI)
All Apple Uatiniosh with a third panr IDDI NuBus card
iosialled, along wiih ihe FOOIIalk proiocols.
Apple Macintosh
Apple Macintosh
Apple Macintosh (LocalTalk)
All Apple Uatifliosb compuiers come equipped wiih locallalk
as standard equipmeoL
Apple Macintosh (RS*422)
Any Apple Macintosh tan use either ol its two serial ports
mS-412) l3t asynchronous services, such as ARAP.
Live Wired
Apple Macintosh
PAP Client
AFP Client/server
Formats
AppleTalk
Token Ring
Apple Macintosh (Token Ring)
An Apple Meciniosli witli a loken Ring NuBus caid inslalled along wlUi ilie Tokenlalk pfoiocols.
AppleTalk Ethemet-to-Ethernet Router
An Applelalk (ouie( thai connecls Iwo (iheinei networks logeHier. (laoiples include Apple's Inlernei flooier (eguipped will) two
[iheinei cards), or even a VAX running Applelalk for VMS over two [ihernet controllers.
AppleTalk Ethernet-to-Token Ring Router
AppleTalk
Ethernet
AppleTalk
Token Ring
AppleTalk Ethemet-to-Token Ring Router
An Applelalk router that connects an Ethernet network and a loken Bing network. Ihe best example ol such a device is Apple's
Internet Router (equipped with an Ethernet card and a loken Ring card).
Appendix A NetPICT Encyclopedia
AppleTalk LocalTalk-to-Ethernet Router
An Appletalk louiet Ihai connecis a IncalTalk network anil an Eiliernei network. Examples incluile Apple's Inieinet nooter (using a
Incallalk connector and an EOieroet card) and dedicated routeis sucti as Stiva's FastPatlt !i.
AppleTalk LocalTalk-to-LocalTalk Router
An Applelalk loutet that connects two localTalk networks. Ihe best example is Apple’s Internet Hooter (using both the printer and
modem ports lor locallalk connections).
AppleTalk LocalTalk-to-Token Ring Router
Ihis diagram represents an Applelalk router that connects a locallalk network and a token Ring network. Ihe best example ol
such a device is Apple's Internet Router (equipped with a locallalk connector and a token Ring card).
Live Wired
Wide-Area Router (l.e. Cisco, Wellfleel)
AppleTalk and
other protocols
Ethernet
AppleTalk and
other protocols
T1 and other
WAN links
AppleTalk multiprotocol WAN router
An Applelalk (mulliproiocol) router that connects an [ttiernet network to a WAN connection (sucli as a II tinkl. Examples include
multiprotocol routers from Dsco, Welllleei and IBM.
PC/WIndows
Banyan VINES
Services
Formats
VINES, AppleTalk
Cabling
Banyan VINES server
A Banyan VINES server running on a PC host. Notice that in addition to the VINES protocol the Applelalk protocol is also installed on
the host. Ihis system also supports the encapsulation ol Applelalk within VINES datagrams.
Cayman GatorBox
Ihe Cayman GatorBox is an AFP-io*NFS gateway. Devices on the AFP side see the GatorBox as an AFP server, while devices on the
NFS side see Ihe GatorBox as iust another NFS client.
Appendix ANetPICT Encyclopedia
Macintosh Macintosh
Dayna NetMounter
Ihe Dayna KeiMounier equips a Maciniosh with ihe Kovell
iranspoii pioiocols, raakinq il appear like any other DOS client
on lire neiwotk. this is an altetnatiye lo lire NetWare lor
MaciniosItptoDuciltont Novell.
DECnet for Macintosh
Part of the PAINWODItS lot Macintosh solution is to provide the
option to turn Macintosh computers into DECnet nodes. DECnet
lor Mac provides E-mail, Die services, DECwindow, and backup
services over locallalk, Elhetnei and RS-422 cabling.
Engage SyncRouter 56K
Engage's Applelalk router, which connects an Ethernet LAN to a switched 561! WAN link. Ibis device works to tandem with another
identical unit.
Engage SyncRouter ISDN
Engage's Applelalk router, which connects an Ethernet IAN to an ISDN WAN link. Ihis device works in tandem with another
identical unit
live Wired
IBM Host
LocalTalk Bridge (Tribe LocalSwitch)
Ihe LocalTalk bridge operaies at Ihe Data link layer. Ii does ool grovide a rooting cagabiliiy. By rapidly switching between devices
on the LocalTalk segmeni the modest bandwidth ol LocalTalk is naiimiied.
LocalTalk Repeater (Farallon PhoneNET Repeater)
The LocalTalk repeater operates at the physical layer by electrically doplicaiing signals Iron one side ol the connection to the other.
It's used to extend the range ol LocalTalk segments.
Appendix A NetPICT Encyclopedia
Macintosh Macintosh
Macintosh IBM 3270 Client
A Maciniosli 3270 leiminal emulatoi running nver a coat taid.
An example nl iliis is lire MaclOMA emulainr and card.
Macintosh
MacTerminal DECnet (CTERM)
overLocalTalk
Here, a Macininsli is used to run a rerminal emulatinn session
using die CIEOM protocol (pan oi ihe DECnel ptolocol) over a
locallall: neivtotk. 01 course. Hie Eihetnei tesidenl host VAX
compuier must be accessible through a EocallaHr lo-Ethernei
router that routes the DECnet protocol.
MacTerminal DECnet (CTERM)
over Ethernet
Here, a Macintosh is used to tun a tetniioal emulation session
with Ihe C1EHM protocol (part ol the DECnet protocol) over an
Eihernei network. Ihe selection of the CIEHM protocol is made
through the Macintosh Communications loolboi.
Macintosh
MacTerminal DECnet (CTERM)
overRS-422
Here, a Macintosh is used to run a terminal emuladon session
using the CIEHM protocol (part ol Ihe DECnel protocol) over a
serial connection through the Mac's RS-422 (printer or modem)
port
Live Wired
Macintosh
MacTerminal Local Area
Transport (LAT) over Ethernet
Here, a Maciniosli is beiao used lo run a lerniinal eiaulaiioa
session using Hie lAI ginincol ever a Etbernei neiwnrk. Ibe
bnsi VAX is also speaking Hie LAT Iranspoit pioincol. The
seleciion of LAI is made through the Macintosh Communica-
tions Toolbox.
Macintosh
MacTerminal Asynchronous
Protocol over RS-422
In Ibis diagram a Maciniosb is used to run a terminal emulation
session using the "async" protocol over an HS-422 connec-
tion (made through the modem or printer port). This is
representative ol most terminal emulators that use async
dial-uplines.
Multiprotocol Ethernet-to-Ethernet Router
AppleTalk,
DECnet, TCP/IP
Ethernet
AppleTalk,
DECnet, TCP/IP
Ethernet
Multiprotocol Ethernet-to-Ethemet Router
A multiprotocol router (AppleTalk, OLCnei and TCP/IP) used to connect two Tihernei segments. Examples include routers burn Cisco
andWelllleet.
Appendix A NetPICT Encyclopedia
Multiprotocol LocalTalk-to-Ethernet Router
AppleTalk,
DECnet, TCP/IP
LocalTalk
AppleTalk,
DECnet, TCP/IP
Ethernet
Multiprotocol LocalTalk-to-Ethemet Router
A fliuliipiotocol rouiei (Applelalk, DECppi, anil ICP/iPI ased lo cnnaeci a Ipcallalk natwoik in an Etheinei sepmeni. Examples
inclpde innieis liom Shiva. Wehsiei, and Cayman.
Macintosh Macintosh
NFS Services on a Macintosh
A Maciniosh equipped wiih Apple's MacICP can support higher
level protocols such as Sun's Network Eiling Sysiem (NFS).
Implemeniaiions el NFS lor the Mat ate provided hy InierCon
and Wollongong.
Novell NetWare for
Macintosh Client
NetWare lor the Mac relies on the Applelalk iranspott to
deliver NetWare services to the desktop. Because it uses the
Applelalk iranspoti any cahling sysiem or router that is
supported by both the Mac and the PC is supported by NetWare
lot the Mac.
Live Wired
PC/Windows
Novell NetWare for
Macintosh Server
lelWare for Macintosh seivices ate imiileiiienieil on a PC
server with a NetWare loarlable Module (NIM). Ihe NLM also
provides the PC wilh Ihe Aoplelallr protocol suite.
Intel-based PC
PC with AppleTalk and a
LocalTalk Card (PhoneNET PC)
fhis PC supports AFP sod PAP protocols with Ihe AppleTalk
protocol Slack over a locallalk cooneciioo. Ihe best example
is Farsllon's PhoneNFI PC product which supports Fthernet
coooectiviiy (in addition to locallalk aod loken Riog).
Intel-based PC
PC with AppleTalk and an
Ethernet Card (PhoneNET PC)
This PC supports AFP and PAP protocols with the Applelalk
protocol slack over an Fthernet connection Ihe host example
is Farsllon's PhoneNET PC product, which supports Ethernet
connectivity (in addition to locallalk aod Token Ring).
Intel-based PC
PC with AppleTalk and a Token
Ring Card (PhoneNET PC)
Fhis PC supports AFP and PAP protocols with Ihe Applelalk
protocol Slack over a Token Ring conneclicn. Again, the best
example is Farsllon's PhoneNFI PC product
Appendix A NetPICT Encyclopedia
Macintosh Running Appie SNA*ps Gateway
Formats
EBCDIC
AppieTalk
SNA
LocalTaik or
Ethernet
Cabiing
SNA*ps Gateway
liiis repfesenis a Maciniosli lunning Apple's SNA*ps paieway. It is used lu conned memlieis on an AppleTalk network lo
anlBMIiosi.
Macintosh
TCP/IP on the Macintosh
Iliis Macinlosli has been equipped with Apple's UacICP, which
puls ihe ItP/IP piolacol slack on the Mac, enabling ii to
pariicipaie in IP network services nver Locallalk or Ethernet.
Macintosh
Telnet Services for the Macintosh
One popular IP service is the Telnet terminal protocol that sits
atop the TCP/IP protocol slack. An etample ol this would be
running the Versalerm Pro terminal emulator over TocalTalk or
EiherneL
Live Wired
Xinet
K-AShare
Sun Workstation
Xinet K-AShare
linet's X'ASliare turns a Sun workstation into an AFP setvet. Notice that the Sun has the AppleTalk ptotocol slack inslalleil.
Xinet K-Spool
Xinei's K-Spool turns a Sun workstation into a PAP print spooler. Ihe spooler accepts Mac ptint jobs, spools then, and sends them
to the appropriate network ptinier.
Listing of
Macintosh
Networking
Companies
3Com Corporation
AESP, Inc.
5400 Baj'front Plaza
1810 NE 144th St.
Santa Clara, CA 95052
North Miami Beach, FL 33181
(800) 638-3266
(800) 446-2377
(408) 764-5000
-A-
(305) 944-7710
AG Group
22540 Camino Diablo
Suite 202
Walnut Creek, CA 94596
Actinet Systems, Inc.
360 Cowper Avenue
(510) 937-7900
Suite 1 1
Alisa Systems, Inc.
Palo Alto, CA 94301
221 East Walnut Street
(415) 326-1321
Suite 175
Pasadena, CA 91101
(818) 792-9474
Live Wired
Andrew Corporation
4301 Westbank Drive
Suite A- 100
Austin, TX 78746
(800) 531-5167
(512) 314-3000
Apple Computer, Inc.
20525 Mariani Avenue
Cupertino, CA 95014
(408) 996-1010
Applied Engineering
P.O. Box 5100
Carrollton, TX 7501 1
(800) 554-6227
APT Communications, Inc.
9607 Dr. Perry Road
Ijamsville, MD 21754
(301) 831-1182
Artisoft Inc.
691 River Road
Tucson, AZ 85704
(602) 293-4000
Asante Technologies
404 Tasman Drive
Sunnyvale, CA 94089
(408) 752-8388
AT&T
295 North Maple Avenue
Basking Ridge, NJ 07920
(908) 221-6153
Avatar Corporation
65 South Street
Hopkinton, MA 01748
(508) 435-3000 East
(408) 727-3270 West
-B-
Banyan Systems, Inc.
120 Flanders Road
Westboro, MA 01581
(508) 898-1000
Blyth Software
1065 East Hillsdale Boulevard
Foster City, CA 94404
(415) 571-0222
Brio Technology, Inc.
444 Castro Street
Suite 700
Mountain View, CA 94041
(415) 961-4110
Appendix B Listing of Macintosh Networking Companies
BT Tymnet
P.O. Box 49019
560 North First Street
San Jose, CA 95161-9019
(408) 922-7583
-c-
Cabletron Systems
35 Industrial Way
Rochester, NH 03867
(603) 332-9400
Cactus Computer, Inc.
1 120 Metrocrest Drive
Suite 103
Carrollton, TX 75006
(214) 416-0525
Caravelle Networks
Corporation
301 Moodie Drive
Suite 306
Nepean, ON 2H 9C4
Canada
(613) 596-2802
Cayman Systems, Inc.
26 Landsdowne Street
Cambridge, MA 02139
(617) 494-1999
cc:Mail, Division of
Lotus Corporation
2141 Landings Drive
Mountain View, CA 94043
(415) 961-8800
CEL Software
P.O. Box 8339
Station F
Edmonton, Alberta T6H 4W6
Canada
(403) 463-9090
Cisco Systems Inc.
1525 O’Brien Drive
Menlo Park, CA 94025
(415) 326-1941
Claris Corporation
5201 Patrick Henry Drive
Santa Clara, CA 95052
(408) 727-8227
Clear Access Corporation
200 West Lowe Street
Fairfield, lA 52556
(515) 472-7077
(800) 522-4252
Codenoll Technology
Corporation
1086 North Broadway
Yonkers, NY 10701
(914) 965-6300
Live Wired
Compatible Systems
Corporation
P.O. Drawer 17220
Boulder, CO 80308
(800) 356-0283
(303) 444-9532
Computer Methods
Corporation
525 Route 73 South
Suite 300
Marlton, NJ 08053
(609) 596-4360
Connectivite Corporation
220 White Plains Road
Tarr)4:own, NY 10591
(914) 631-5365
CSG Technologies
530 William Penn Place
Suite 329
Pittsburgh, PA 15219
(412) 471-7170
(800) 366-4622
-D-
Data Spec
9410 Owensmouth Avenue
Chatsvvorth, CA 9131 1
(818) 772-9977
DataViz
55 Corporate Drive
Trumbull, CT 06611
(203) 268-0030
Dayna Communications, Inc.
50 South Main Street
Fifth Floor
Salt Lake City, UT 84144
(801)531-0203
Digital Communications
Association, Inc.
1000 Alderman Dr.
Alpharetta, GA 30202
(404) 442-4000
Digital Equipment
Corporation
146 Main Street
Maynard, MA 01754
(508) 493-5111
Digital Products Inc.
4 1 1 Waverley Oaks Road
Waltham, MA 02154
(617) 647-1234
Appendix B Listing of Macintosh Networking Companies
-E-
EDI Communications
Corporation
20440 Town Center Lane
Suite 4E1
Cupertino, CA 95014
(408) 996-1343
EMAC Division of Everex
Systems, Inc.
48431 Milmont Drive
Fremont, CA 94538
(800) 628-3837
Engage, Inc.
9053 Soquel Drive
Aptos, CA 95003
(408) 688-1021
Everywhere Development
Corporation
2176 Torquay Mews
Mississaugua, ON LSN 2M6
Canada
(416) 819-1173
-F-
Fairfield Software, Inc.
200 West Lowe Street
Fairfield, lA 52556
(515) 472-7077
Farallon Computing, Inc.
2000 Powell Street
Suite 600
Emeryville, CA 94608
(800) 998-7761
(510) 596-9100
Focus, Inc.
800 West Cummings Park
Suite 4500
Woburn, MAO 1801
(800) 538-8866
(617) 938-8088
-H-
Hayes Microcomputer
Products, Inc.
P.O. Box 105203
Atlanta, GA 30348
(404) 840-9200
Live Wired
Helios USA
10601 South DeAnza Blvd.
Suite 103
Cupertino, CA 95014
(408) 864-0690
Hewlett-Packard Co.
5301 Stevens Creed Boulevard
Santa Clara, CA 95052
(800) 752-0900
IBM Corporation
1 133 Westchester Avenue
White Plains, NY 10604
(800) 426-3333
IDEAssociates, Inc.
29 Dunham Road
Billerica, MA 01821
(508) 663-6878
impulse Technology
210 Dahlonega Street
Suite 206
Cumming, GA 30130
(404) 889-8294
Infotek, Inc.
56 Camille Lane
East Patchogue, NY 1 1772
(516) 289-9682
InterCon Systems Corp.
950 Herndon Parkway
Suite 420
Herndon, VA 22070
(703) 709-9890
international Transware, inc.
1503 Grand Road
Suite 155
Mountain View, CA 94040
(800) 999-6387
(415) 903-2300
IPT - Information
Presentation Technologies
555 Chorro Street
San Luis Obispo, CA 93405
(805) 541-3000
-K-
Kandu Software Corp.
2305 North Kentucky Street
Arlington, VA 22205
(703) 532-0213
Appendix B Listing of Macintosh Netu'orking Companies
Keywork Technologies
2816 Eleventh Street, NE
Calgary, Alberta T2E 7S7
Canada
(403) 250-1770
Marietta Systems
International
29 El Cerrito Avenue
San Mateo, CA 94402
(415) 344-1519
MCI
nil 19th Street NW
Suite 500
Washington, DC 20036
(800) 444-6245
MDG Computer Services
634 South Dunton
Arlington Heights, IL
60005-2544
(708) 818-9991
(708) 453-6330
Microsoft Corporation
One Microsoft Way
Redmond, WA 98052-6399
(206) 882-8080
Miramar Systems
201 North Salsipuedes
Suite 204
Santa Barbara, CA93103
(805) 965-5161
Mitem Corporation
2105 Hamilton Avenue
Suite 350
San Jose, CA 95125
(408) 559-8801
Motorola Computer Group
10700 North De Anza Boulevard
Cupertino, CA 95014
(800) 759-1107 Ext. MU
MultiAccess Computing
Corporation
5350 Hollister Avenue
Suite C
Santa Barbara, CA 931 1 1
(805) 964-2332
National Semiconductor
Corporation
2900 Semiconductor Drive
P.O. Box 58090
Santa Clara, CA 95052-8090
(408)721-5000
Live Wired
Neon Software, Inc.
1009 Oak Hill Road
Suite 203
Lafayette, CA 94549
(510) 283-9771
Network General Corporation
4200 Bohannon Drive
Menlo Park, CA 94025
(415) 688-2700
Network Resources
Corporation
736 South Hillview Drive
Milpitas, CA 95035
(408) 263-8100
Nevada Western
615 North Tasman Drive
Sunnyvale, CA 94089-1950
(408) 734-8727
Novell, Inc.
122 East 1700 South
Provo, UT 84606
(800) 638-9273
(801) 379-5900
- 0 -
ON Technology, Inc.
155 Second Street
Cambridge, MA 02141
(617) 876-0900
Oracle Corporation
500 Oracle Parkway
Redwood Shores, CA 94065
(415) 506-7000
-P-
Pacer Software, Inc.
7911 Herschel Avenue
Suite 402
La Jolla, CA 92037
(619) 454-0565
Photonics Corporation
200 East Hacienda Avenue
Campbell, CA 95008
(408) 370-3033
Appendix B Listing of Macintosh Networking Companies
-R-
RacaMnterian, Inc.
155 Swanson Road
Foxborough, MA 01719
(508) 263-9929
-s-
Shiva Corporation
One Cambridge Center
Cambridge, MA 02142
(800) 458-3550
(617) 252-6300
Sitka Corporation
950 Marina Village Parkway
Alameda, CA 94501
(800) 445-8677
(510) 769-9669
Soft-Switch Inc.
640 Lee Road
Wayne, PA 19087-5698
(215) 640-9600
SoftWriters, Inc.
P.O. Box 1308
Round Rock, TX 78680
(512) 244-3999
Sonic Systems. Inc.
333 West El Camino Real
Suite 280
Sunnyvale, CA 94087
(800) 535-0725
(408) 736-1900
Standard Microsystems
Corporation
80 Arkay Drive
Hauppauge, NY 11788
(516) 273-3100
StarNine Technologies. Inc.
2126 Sixth Street
Berkeley, CA 94710
(510) 548-0391
Synergy Software
2457 Perkiomen Avenue
Mt. Penn, PA 19606
(215) 779-0522
-T-
Talaris Systems. Inc.
6059 Cornerstone Court West
San Diego, CA 92121
(619) 587-0787
Live Wired
TechGnosis Inc.
301 Yamato Road
Suite 2200
Boca Raton, FL 33431
(407) 997-6687
TechWorks
4030 Broker Lane West
Suite 350
Austin, TX 78759
(800) 879-7745
(512) 794-8533
Terranetics
1538 North Martel Avenue
Suite 413
Los Angeles, CA 90046
(818) 446-7692
Thomas-Conrad Corporation
1908-R Kramer Lane
Austin, TX 78758
(800) 332-8683
(512) 836-1935
Thursby Software
Systems, Inc.
5840 West Interstate 20
Suite 100
Arlington, TX 76017
(817) 478-5070
Tribe Computer Works
1195 Park Avenue
Suite 211
Emerj'ville, CA 94608
(510) 547-3874
Trik, Inc.
400 West Cummings Park
Suite 2350
Woburn, MAO 1801
(800) 766-0356
(617) 933-8810
-u-
Ungermann-Bass, Inc.
3990 Freedom Circle
P.O. Box 58030
Santa Clara, CA 95052
(800) 873-6381
(408) 496-0111
United Data Corporation
3755 Balboa Street
Suite 203
San Francisco, CA 94121
(415) 221-8931
Appendix B Listing of Macintosh Networking Companies
US Sprint Communications
P.O. Box 84 17
1200 Main St., 4th Floor
Kansas City, MO 64105
(800) 877-2000
-W-
Webster Computer
Corporation
2109 O’Toole Avenue
Suite J
San Jose, CA 95131
(800) 457-0903
(408) 954-8054
White Pine Software
40 Simon Street
Suite 201
Nashua, NH 03060-3043
(603) 886-9050
The Wollongong Group, Inc.
1 129 San Antonio Road
Palo Alto, CA 94303
(415) 962-7100
WordPerfect Corporation
1555 North Technology Way
Orem, UT 84057
(801) 225-5000
(800) 451-5151
-X-
Xinet
2560 9th Street
Suite 312
Berkeley, CA 94710
(510) 845-0555
WilTel
P.O. Box 21348
Tulsa, OK 74121
(800) 642-2299
Live Wired
A Guide to Networking Macs
Index
Liue Wired
Symbols
* (asterisk) wildcard character,
with Network Visible Entities
(NVEs), 158
= (equal sign) wildcard
character, with Network
Visible Entities (NVEs), 158
1-2-3 (Lotus), 212
Data Access Language
(DAL), 109
format support, 34-35
lOBase-2 (thinwire) Ethernet,
92, 192-193, 195
lOBase-5 (thickwire) Ethernet,
93, 190-192, 196
lOBase-T (twisted-pair)
Etliernet, 68, 95, 193-194, 196
single star networks,
299-300
lOBroad-36 Ethernet, 192
3270 terminal emulators, 106,
269-270
3Com Corporation, 383
EtherLink series, 228
5250 terminal emulators,
106, 270
8- pin D1N8 connectors, 181
802.3 (lOBase-F) IEEE
standard, 289
9- pin DB9 connectors, 181
A
A/UX (Apple), DAL servers,
109, 234
AARP (AppleTalk Address
Resolution Protocol), 55,
141-142
ACCUNET system (AT&T), 204
acknowledge control
packets, 139
Acrobat (Adobe), 123
Actinet Systems, Inc., 383
ARCTalk cards, 229
active star topologies, 183-185
single LocalTalk networks,
298-299
Address Mapping Table
(AMT), 55, 141
Address Resolution Protocol
(ARP), 55
addressing, 49-50
in AppleTalk, 137-147
logical, 50-53
physical, 54-55
TCP/IP, 245-247
Adobe Systems
Acrobat, 123
PostScript, 1 19
ADSP (Apple Data Stream
Protocol), 132
Advanced Peer-to-Peer
Networking (APPN), 274
Advanced Program-to-
Program Communications
(APPC), 274
Advanced Software Concepts
(ASC)
asc3270, 241,270
asc5250, 241
FTPShare, 236
terminal emulation
products, 241-242
Index
AEP (AppleTalk Echo
Protocol), 134
AESP, Inc., 383
AFP, see Apple Filing Protocol
AG Group, 383
EtherPeek, 322-323
LocalPeek, 322-323
TokenPeek, 322-323
agent programs, 314
AIR (Apple Internet Router),
172-174, 203
Aldus PageMaker, translating
formats between versions,
121
Alisa Systems, Inc., 383
AlisaShare, 256
MailMate QM, 112
AlisaShare (Alisa Systems), 256
All-In- 1 Mail for Macintosh
(DEC), 88, 256
mail gateways, 112
Altair II wireless networks
(Motorola), 201-202
AMP (Apple Management
Protocol), 315
amplitude, modulating, 58
AMT (Address Mapping
Table), 55, 141
analog dial-up, 203-204
integrating Macintosh and
DEC VAX, 267
Andrew Corporation, 384
Andyne’s Graphical Query
Language (GQL)
Data Access Language
(DAL), 109
Anonymous File Transfer
Protocol (FTP), 235
ANSI X3T9.5: FDDI
standard, 290
AOCE (Apple Open Collabora-
tion Environment), 105-106
APPC (Advanced Program-to-
Program Communications),
274
Apple Computer, Inc., 384
Apple Internet Router (AIR),
172-174, 203
Apple Open Collaboration
Environment (AOCE),
105-106
AppleShare, 99-101
AppleTalk, see AppleTalk
AppleTalk/ IP Wide Area
Extension, 174
AppleTalk/X.25 Wide Area
Extension, 173
Attachment Unit Interface
(AUI) connectors, 194-195
Data Access Language
(DAL), see Data Access
Language
Easy Open file translator, 126
Ethernet cards, 194-195
EtherTalk, 26-27
FDDI (Fiber Distributed
Data Interconnect), 28-29
future, 337-339
Inter«Poll, 318-319
LaserWriter printers, 104
LocalTalk, seeLocalTalk
Macintosh coax and twdnax
cards, 276
MacODA, 122-123
MacTCP, 23-24, 226, 244-247
MacX, 240
Live Wired
MPW, ASCII text. 116
Newton, 29
NuBus Token Ring card, 229
Printer Access Protocol
(PAP). 103-105, 131-132
QuickTime for Windows, 212
Serial NB card, 277
SNA»ps
3270 terminal emulator,
270
gateway, 274, 277, 381
TokenTalk, 27
Apple Data Stream Protocol
(ADSP), 132
Apple Filing Protocol
(AFP), 129
AppleShare, 99-101
gateways with Network
Filing System (NFS), 87-88
Macintosh File Sharing
(System 7). 101-102
on UNIX computers, 232-233
servers
for other platforms,
102-103
NetPICT diagrams,
369-370
VAXshare, 253-256
Apple Internet Router (AIR),
172-174,203
Apple Management Protocol
(AMP), 315
Apple Open Collaboration
Environment (AOCE),
105-106
Apple Remote Access (ARA),
AppleTalk/LAT Gateway, 265
Apple Secure Data Stream
Protocol (ASDSP), 106,
132, 323
Apple Transaction Protocol
(ATP), 133-134
Apple Update Routing Proto-
col (AURP), 133,171-174
AppleShare (Apple), 99-101
AppleShare icon, 100
AppleTalk (Apple), 23
AppleTalk Address Resolu-
tion Protocol (AARP), 55
cabling support, 26-29
Chooser, see Chooser
gateways, 87-90
logical addressing, 50-53
names
Network Visible Entities
(NVEs), 157-165
zone, 142-147
NetPICT diagrams,
372-374, 380
nodes, turning PCs into,
219-222
numbers
network, 137-138
node, 138-142
Phase 1 networks
broadcasting versus
Phase 2 multicasting,
153-155
device limits, 148-149
network number
assignments, 149-152
Phase 2 networks
network number
assignments, 149-152
Index
transition routing,
152-153
phase development, 148
routers, 51, 75-79, 166
tunneling protocols,
174-178
updating routing tables
continuously, 166-170
updating routing tables
periodically, 171-174
subdividing networks, 81-82
suite of protocols, 127
Data Link OSI layer,
135-136
Network OSI layer,
134-135
Presentation OSI
layer, 129
Session OSI layer, 129-132
Transport OSI layer,
132-134
AppleTalk Address Resolution
Protocol (AARP), 55, 141-142
AppleTalk Echo Protocol
(AEP), 134
AppleTalk for VMS, 260
AppleTalk LAT Gateway, 88-89
AppleTalk Remote Access
Protocol (ARAP), 27-28
AppleTalk Session Protocol
(ASP), 131
AppleTalk/DECnet Gateway,
88, 263-264
AppleTalk/DECnet MCT
tool, 264
AppleTalk/IP Wide Area
Extension, 174
AppleTalk/LAT Gateway,
264-265
AppleTalk/X.25 Wide Area
Extension, 173
Application OSI layer, 42
with Macintosh, 44
Applied Engineering, 384
APPN (Advanced Peer-to-Peer
Networking), 274
APT Communicadons, Inc., 384
ARA (Apple Remote Access),
AppleTalk/LAT Gateway, 265
ARAP (AppleTalk Remote
Access Protocol), 27-28
ARCNET, 200
integrating Macs and PCs,
229-230
networking cards, 36
ARCTalk cards (ACTINET
Systems), 229
ARP (Address Resolution
Protocol), 55
Artisoft Inc., 384
AS/400 computers (IBM),
integrating with Macintosh
cabling systems
direct connections,
276-277
Ethernet, 278
Token Ring, 278
formats, 273
services
DAL database, 272
terminal, 269-272
transport protocols, 273-275
Asante Technologies, 384
AsantePrint bridge, 187
Live Wired
ASC, see Advanced Software
Concepts
asc3270 (Advanced Software
Concepts), 241, 270
asc5250 (Advanced Software
Concepts), 241
ASCII (American Standard
Code for Information
Interchange), 20-21
format, 116
integrating Macintosh
and DEC VAX, 258
integrating Macs and
PCs, 214
ASDSP (Apple Secure Data
Stream Protocol), 106,
132, 323
ASP (AppleTalk Session
Protocol), 131
asterisk (*) wildcard character,
with Network Visible Entities
(NVEs), 158
Asynchronous protocol,
NetPICT diagrams, 378
AsyncServeR (Computer
Methods Corporation), 28,
267, 335
AT&T, 384
ACCUNET system, 204
ATP (Apple Transaction
Protocol), 133-134
Attachment Unit Interface
(AUI) connectors, 194-195
AURP (Apple Update Routing
Protocol), 133, 171-174
AutoCAD (Autodesk), 212
DXF (Drawing Exchange
Format) format, 217-218
Avatar Corporation, 384
Macintosh coax and twinax
cards, 276
Macintosh/SNA
gateways, 277
MacMainFrame, 270
B
backbone (bus) topologies,
76-77, 92-93
bridged, multiple LocalTalk
networks, 305-306
Ethernet, 196
multiple LocalTalk
networks, 306-307
FDDI
multiple Ethernet
networks, 309
multiple Ethernet WAN
networks, 309-311
PhoneNET, 185
routed
multiple Ethernet
networks, 307-308
multiple LocalTalk
networks, 304-305
bandwidth, comparing
Ethernet and LocalTalk,
330-331
Banyan Systems, Inc., 384
VINES, 35, 224
NetPICT diagrams, 374
baseband, 59, 192
Bellman-Ford routing, 169-170
binary data, 19
Index
binary formats, 121
integrating Macintosh and
DEC VAX, 258-259
bitmap graphics format, 118
bits, 19
Black Box Corporation, 292
Blacksmith (CEIJ, 272
Blyth Software, 384
BNC (twist and lock)
connectors, 192
Both (Connectivite), 272
bridges, 69-71
backbone, multiple
LocalTalk networks,
305-306
Ethernet, 71-72
filtering, 70
for LAT networks, 83-86
isolating high-traffic nodes,
332-335
LocalTalk, 71
NetPICT diagrams, 376
Source Routing Bridges
(SRBs), 334
star, single LocalTalk
networks, 302
TribeStar (Tribe), 305
Brio Technology, Inc., 384
broadband networking, 59
broadcasting. Phase 1.
versus Phase 2 multicasting,
153-155
brouters, 85
BT Tymnet, 385
bus topologies, .see backbone
topologies
bytes, 19
c
cable
anticipating wiring
standards, 194
coaxial (coax), 276
fiber optic, 289-290
twinaxial (iwinax), 276
Unshielded Twisted-Pair
(UTP), 289
Cabletron Systems, 385
Cabling NetPICT layer, 25-29
ARCNET, 200
conventional analog dial-
up, 203-204
designing, 284-285
strategy guidelines,
288-292
Ethernet, 187-196
FDDI (Fiber Distributed
Data Interface), 197-199
integrating Macintosh
and DEC VAX, 266-268
and IBM mainframes/
midranges, 276-278
and PCs, 227-230
and UNIX, 248
ISDN (Integrated Services
Digital Network), 204-205
leased-line services, 205
LocalTalk, 179-187
multiple Ethernet
FDDI backbone, 309
FDDI backbone WAN,
309-311
routed backbone and
star, 307-308
Live Wired
multiple LocalTalk
bridged backbone,
305- 306
Ethernet backbone,
306- 307
routed backbone, 304-305
serially-routed, 302-304
packet switched, 207
PhoneNET (Farallon),
182-187
Serial RS-232/422, 199-200
single Ethernet
daisy-chain, 294-296
twisted-pair star, 299-300
single LocalTalk
active star, 298-299
bridged star, 302
daisy-chain, 292-294
multiple star, 300-301
passive star, 297-298
standard Macintosh, 33
standard PC, 35-36
structured wiring imple-
mentations, 311-312
Switched 56K Service, 204
Token Ring, 196-197
wireless, 200-202
X.25 packet switched,
206-207
Cactus Computer, Inc., 385
Caravelle Networks
Corporation, 385
cards, 35-36
Ethernet, 54
Apple, 194-195
cost, 187, 189
FDDI, cost, 198
Macintosh coax and
twinax, 276
Serial NB (Apple), 277
Token Ring, cost, 197
Carrier Sense Multiple Access/
Collision Avoidance (CSMA/
CA),63
Carrier Sense Multiple Access/
Collision Detection (CSMA/
CD), 62-63
Category 5 UTP, 289
Cayman Systems, Inc., 385
GatorBox gateway, NetPICT
diagrams, 374
cc:Mail, Division of Lotus
Corporation, 385
CE Software’s QuickMail,
111-112
CEL Software, 385
Blacksmith, 272
Chooser, 155-156
determining node
addresses, 156-157
indicating network
problems, 318
standard Service layer, 32
troubleshooting, 324-330
circuit-switched networks,
47-50
Cisco Systems Inc., 385
multiprotocol routers, 205
routers, 85, 336
Claris Corporation, 385
ClarisWorks, 212
FileMaker Pro, 212
XTND data format
translators, 117, 124-125
Index
Clear Access (Fairfield
Software), 109
Clear Access Corporation, 385
client/server computing, 39
Apple Filing Protocol (AFP)
AppleShare, 99-101
for other platforms.
102-103
Macintosh File Sharing
(System 7), 101-102
Data Access Language
(DAL), 108-110
Printer Access Protocol
(PAP), 103-105
X- Window standard, 238-241
client/server model, 18
coaxial (coax) cable, 276
Codenoll Technology
Corporation, 385
FDDI cards, 198
collisions, 60
common bus topologies, see
backbone topologies
communication layers
Expression, 10
Idea, 10
matching, 13-15
Medium, 11-12
Transport, 11
Compatible Systems
Corporation, 386
EtherWrite bridge, 187
routers, 335
compound documents, 122
Computer Methods
Corporation, 386
AsyncServeR, 28, 267, 335
concentrators, 90
configuring nodes, 313-315
Connectivity Corporation, 386
Both, 272
connectors
8- pin DIN8, 181
9- pin DB9, 181
Attachment Unit Interface
(AUl), 194-195
BNC (twist and lock), 192
RJ-11, 182
S>mchronous Data Link
Control (SDLC), 275
tee, 193
transceivers, 191
conventional analog dial-up
networks, 203-204
converting formats, 124-126
Cooperative Printing
Solutions Inc. (COPS Inc.),
turning PCs into AppleTalk
nodes, 222
cost
Altair II wireless networks
(Motorola), 202
Ethernet cards, 187, 189
FDDI cards, 198
multiprotocol routers, 205
single Ethernet daisy-chain
networks, 295-296
single LocalTalk active star
networks, 298-299
single LocalTalk daisy-
chain networks, 293
single LocalTalk multiple
star networks, 301
Live Wired
single LocalTaJk passive star
networks, 297
Token Ring cards, 197
X.25 packet switched
networks, 206-207
CSG Technologies, 386
CSMA/CA (Carrier Sense
Multiple Access/ Collision
Avoidance), 63
CSMA/CD (Carrier Sense
Multiple Access/Collision
Detection), 62-63
CTERM Communications
Tool, 262
CTERM protocol, 262
NetPICT diagrams, 377
cyclic redundancy checks, 57
D
daisy-chain topologies, 67,
91-92
single Ethernet networks,
294-296
single LocalTalk networks,
292-294
data
binary, 19
formats, see formats
Data Access Language (DAL)
(Apple), 108-110
for VMS, 256
integrating Macintosh
and UNIX, 234
and IBM mainframes/
midranges, 272
Data Link OSI layer, 41-43
Macintosh protocols,
135-136
with Macintosh, 45
Data Spec, 386
databases, relational
accessing, 108-110
see also Data Access
Language (DAL)
DataClub (Novell), 224
Datagram Delivery Protocol
(DDP), 134-135
Short and Long, 138
datagrams, 43
Data Viz, 386
MacLinkPlus, 117,214
Dayna Communications,
Inc., 386
EtherPrint bridge, 186, 189
EtherPrint Plus bridge, 187
NetMounter, 223
NetPICT diagrams, 375
Network Vital Signs, 331-332
DB2 databases (IBM),
accessing, 109
DB9 connectors, 181
DBC/ 1012 databases
(Teradata), accessing, 109
DCA, see Digital Communica-
tions Association, Inc.
DDP (Datagram Delivery
Protocol), 134-135, 138
DEC (Digital Equipment
Corporation), 386
All-In-1 Mail for Macintosh,
88, 1 12, 256
DAL servers, 109
Index
DECnet, see DECnet
DECwindows, 238, 252-253
Local Area Transport (LAT),
82-86
multiprotocol routers, 205
PATHWORKS for Macintosh,
88-89, 251-257
Rdb databases, accessing,
109
Ultrix, 233-234
VAX, see VAX/ VMS
VT-series terminals,
emulating, 106
DECnet (DEC), 23-24, 261-262
AppleTalk/ DECnet
Gateway, 263-264
integrating Macs and PCs,
224-225
NetPlCT diagrams, 375
subdividing networks, 81-82
DECservers, 262
DECwindows (DEC), 238
MacX, 252-253
Default zones, 145
designing networks
by layers, 283-285
cabling-system strategy
guidelines, 288-292
multiple Ethernet
FDDl backbone, 309
FDDI backbone WAN,
309-311
routed backbone and
star, 307-308
multiple LocalTalk
bridged backbone,
305-306
Ethernet backbone,
306-307
routed backbone,
304-305
serially-routed, 302-304
single Ethernet
daisy-chain, 294-296
twisted-pair star, 299-300
single LocalTalk
active star, 298-299
bridged star, 302
daisy-chain, 292-294
multiple star, 300-301
passive star, 297-298
structured wiring imple-
mentations, 311-312
witii NetPlCT sjnnbols,
285-287
dial-up networks, 203-204
integrating Macintosh and
DEC VAX, 267
Digital Communications
Association, Inc. (DCA), 386
IRMA Workstation for
Macintosh, 270
Macintosh coax and twinax
cards, 276
Macintosh/SNA gateways,
277
Digital Equipment
Corporation, see DEC
Digital Products Inc., 386
SprintTALK bridge, 187
digital signatures, 106
D1N8 connectors, 181
Dls (domain headers), 177-178
Live Wired
documents
compound, 122
interchanging, 122
common format stan-
dards versus format
conversion, 124
with Adobe Acrobat, 123
with Aldus PageMaker,
121
with Apple Easy Open,
126
with Apple MacODA,
122-123
with Claris XTND, 117,
124-125
with Microsoft Word, 117
domain headers (DIs), 177-178
dotted quads, 246
Drawing Exchange Format
(DXF), 217-218
drivers
ELAP software, 190
FDDlTalk, 198
dynamic node addressing,
138-142
E
E-mail (electronic mail),
110-113
PATHWORKS for Macintosh
(DEC), 256-257
Easy Open (Apple) file
translator, 126
EBCDIC (Extended Binary
Coded Decimal Interchange
Code), 273
EDI Communications
Corporation, 387
EIA/TIA 568 and TSB-36
standard, 289
ELAP, see EtherTalk Link
Access Protocol
EMAC Division of Evcrcx
Systems, Inc., 387
EMBARC (Motorola), 201
emulators, see terminal
emulators
Encapsulated PostScript
(EPSF), 119-120
encapsulating protocols,
174-178
encrypting network traffic, 106
Engage Communications,
Inc., 387
ISDN AppleTalk routers, 205
SjmcRouter
LTNT router, 204
NetPICT diagrams, 375
enquiry control packets, 139
EPSF (Encapsulated
PostScript), 119-120
equal sign (=) wildcard charac-
ter, with Network Visible
Entities (NVEs), 158
errors, cyclic redundancy
checks, 57
EtherLink series (3Com), 228
Ethernet, 187-190
lOBroad-36, 192
Apple cards, 194-195
AppleTalk support for, 26-27
backbone, multiple Local-
Talk networks, 306-307
Index
bandwidth
capacity, 66
comparing with
LocalTalk, 330-331
bridges, 70-72
broadcasts, 84
Carrier Sense Multiple
Access/Collision Detec-
tion (CSMA/CD), 62-63
design considerations,
195-196
integrating Macintosh
and DEC VAX, 267
and IBM mainframes/
midranges, 278
and PCs, 228
and UNIX, 248
isolating high-traffic nodes,
332-335
multiple
FDDl backbone, 309
FDDl backbone WAN,
309-311
routed backbone and
star, 307-308
NetPICT diagrams, 369-373,
377-380
network signals, 60-61
networking cards, 36
physical addressing, 54-55
repeaters, 68
single
daisy-chain, 294-296
twisted-pair star, 299-300
star topologies, 94-95
thickwire (lOBase-5), 93,
190-192, 196
thinwire (lOBase-2), 92,
192-193, 195
twisted-pair (lOBase-T), 68,
95, 193-194, 196
zone names, 144-147
EtherPeek (AG Group), 322-323
EtherPrint bridge (Dayna
Communication), 186, 189
EtherPrint Plus bridge (Dayna
Communication), 187
EtherTalk (Apple), 26-27,
188-189
establishing node
numbers, 141
EtherTalk Link Access Protocol
(ELAP), 136
software drivers, 190
EtherWrite bridge (Compat-
ible Systems), 187
Everywhere Development
Corporation, 387
Excel (Microsoft), 212
Data Access Language
(DAL), 109
Executive Workstation
(MediaWorks), 271
eXodus (Wliite Pine
Software), 240
Expression layer of
communications, 10
matching senders and
receivers, 13-15
Extended Binary Coded
Decimal Interchange Code
(EBCDIC), 273
exterior routers, 176
live Wired
F
Fairfield Software, Inc., 387
Clear Access, 109
Farallon Computing, Inc., 387
LocalPath bridge, 187
NetAtlas, 320
PhoneNET, 182-187
NetPICT diagrams,
376, 380
PhoneNET PC, 219-221
PhoneNET Repeater, 67
PhoneNET Talk PC, 23
StarController repeater, 67,
94, 183, 185
FastPath router (Shiva), 80-81
FDDI (Fiber Distributed Data
Interconnect) (Apple), 28-29,
197-199
backbone topologies
multiple Ethernet
networks, 309
multiple Ethernet WAN
networks, 309-311
NetPICT diagrams, 369, 371
ring topologies, 95
wiring standards, 194
FDDITalk drivers, 198
fiber optic cabling, 289-290
File Transfer Protocol (FTP),
integrating Macintosh and
UNIX, 235-236
FileMaker Pro (Claris), 212
files
for NetPICT symbols, 286
names, with mixed-plat-
form environments, 103
sharing
Apple Filing Protocol
(AFP), on other
platforms, 102-103
AppleShare, 99-101
Macintosh File Sharing
(System 7), 101-102
filtering bridges, 70
Flight Simulator (Microsoft),
212
Focus, Inc., 387
TurboStar repeater, 68,
94, 185
Format NetPICT layer, 19-22
designing, 284-285
integrating Macintosh
and DEC VAX, 258-259
and IBM mainframes/
midranges, 273
and PCs, 214-218
and UNIX, 242-243
standard Macintosh, 32-33
standard PC, 34-35
formats
ASCII, 116,214, 258
binary, 121, 258-259
common standard, 122-123
versus conversion, 124
EPSF (Encapsulated
PostScript), 119-120
GIF (Graphic Interchange
Format), 120-121
graphics, 118-120,215-218
MacPaint, 118
ODA (Open Document
Architecture), 122-123
PDF (Portable Document
Format), 123
Index
PICT, 118-119
PICT2, 118
platform differences,
115-116
PostScript, 119
raster, 120-121
TIFF (Tagged Image File
Format), 120-121
word processing, 1 16-117,
214-215
Frame Relay packet switched
networks, 207
frames, 43, 56-57
Frequency Modulation, 58
front-end to terminal-based
services, 107
FTP (File Transfer Protocol),
integrating Macintosh and
UNIX, 235-236
FTP Software’s PC/TCP
Plus, 226
FTPShare (Advanced Software
Concepts), 236
G
gateways, 86-87
AppleTalk, 87-90
AppleTalk/ DCCnet, 263-264
AppleTalk/ LAT, 264-265
Macintosh/SNA, 277
mail, 111-112
NetPICT diagrams, 374, 381
SNA»ps (Apple), 274, 277
GatorBox gateway (Cayman),
NetPICT diagrams, 374
Graphic Interchange Format
(GIF), 120-121
Graphical Query Language
(GQL) (Andyne), Data Access
Language, 109
graphics formats
EPSF (Encapsulated
PostScript), 119-120
integrating Macs and PCs,
215-218
MacPaint bitmap, 118
PICT, 118-119
PICT2, 118
PostScript, 119
H
Hayes Microcomputer
Products, Inc., 387
Helios USA, 388
Hewlett-Packard Co., 388
PCL format, 35
UNIX computers
/VFP/PAP services,
232-233
DAL servers, 109, 234
Hijaak (Inset Systems), 217
hops, 76
hubs, 90
HyperCard
ASCII text, 116
creating custom terminal
front ends, 272
DAL client applications, 108
HyperFTP freeware stack, 235
live Wired
i-j
IBM Corporation, 388
3270 and 5250 terminals,
emulating, 106
DAL servers, 109
DB2 databases, accessing,
109
mainframes/ midranges,
integrating with
Macintosh, 269-278
multiprotocol routers, 205
NetPICT diagrams, 376-377
PROFS mail gateways, 112
icons, AppleShare, 100
Idea layer of communication,
10
matching senders and
receivers, 13-15
IDEAssociates, Inc., 388
IEEE 802.3 (lOBase-F)
standard, 289
IGES (Initial Graphics Ex-
change Specification) format,
217-218
Impulse Technology, 388
FDDI cards, 198
Informix databases,
accessing, 109
Infotek, Inc., 388
Ingres databases, accessing,
109-110
Initial Graphics Exchange
Specification (IGES) format,
217-218
Inset Systems’ Hijaak, 217
Integrated Services Digital
Network (ISDN), 204-205
Inter»Poll (Apple), 318-319
InterCon Systems Corp., 388
InterPrint, 233
NFS, 226
NFS/Share, 237
Planet X, 240
WatchTower, 317
interior routers, 176
International Transware,
Inc., 388
Internet Protocol (IP), see
TCP/IP
internetworks, 51, 81
InterPrint (InterCon), 233
IP addressing, 245-247
IPT (Information Presentation
Technologies), 388
Partner, 237
IRMA Workstation for
Macintosh (DCA), 270
ISDN (Integrated Ser\ices
Digital Network), 204-205
ISO standards. Open
Document Architecture
(ODA), 122-123
K
K-AShare (Xinet), 232
NetPICT diagrams, 382
K-Spool (Xinet), 232-233
NetPICT diagrams, 382
K-Talk (Xinet), 244
Kandu Software Corp., 388
Index
Keywork Technologies, 389
Keypak, 259
Kinetics FastPath router, 75
L
LAN Manager (Microsoft), 35
LanRover (Shiva), 28, 335
LANs, see Local Area Networks
LANsurveyor (Neon Software),
320
LaserWriter printers, 104
NetPICT diagrams, 370
virtual, 233
LAT, see Local Area Transport
layers
communication
Expression, 10
idea, 10
matching, 13-15
Medium, 1 1-12
Transport, 11
NetPICT
Cabling, see Cabling
NetPICT layer
Format, see Format
NetPICT layer
matching, 37-38
Transport, see Transport
NetPICT layer
Service, see Service
NetPICT layer
standard Macintosh,
31-33
standard PC, 34-36
swapping, 36-37
OSl Reference Model, 40-43
with Macintosh, 44-45
leased-Iine services for
WANs, 205
Lightspeeed C (Symantec),
ASCII text, 116
LLAP (Local Falk Link Access
Protocol), 136
Local Area Netw'orks (LANs),
65-66
ARCNET, 200, 229-230
Ethernet, see Ethernet
FDDI,5eeFDDI
LocalTalk, seeLocalTalk
PhoneNET, 182-187
Serial RS-232/422, 199-200
Token Ring, see Token Ring
wireless, 200-202
Local Area Transport (LAT)
(DEC), 82-86
AppleTalk/LAT Gateway,
264-265
integrating Macintosh and
DEC VAX, 262-263
NetPICT diagrams, 378
LocalPath bridge (Farallon),
187
LocalPeek (AG Group), 322-323
LocalSwitch bridge (Tribe),
71,302
NetPICT diagrams, 376
LocalTalk (Apple), 26, 179-182
AppleTalk LAT Gateway,
88-89
bandwidth, comparing with
Ethernet, 330-331
bridges, 71
Live Wired
Carrier Sense Multiple
Access/Collision Avoid-
ance (CSMA/CA), 63
daisy-chain topologies, 91
integrating Macintosh
and DEC VAX, 266
and PCs, 227
and UNIX, 248
isolating high-traffic nodes,
332-335
multiple
bridged backbone,
305- 306
Ethernet backbone,
306- 307
routed backbone, 304-305
serially-routed, 302-304
NetPICT diagrams, 370-371,
373, 376-377, 379-380
network signals, 62-63
repeaters, 67-68
single
active star, 298-299
bridged star, 302
daisy-chain, 292-294
multiple star, 300-301
passive star, 297-298
star topologies, 94
zone names, 143-144
LocalTalk Link Access Protocol
(LEAP), 136
logical addressing, 50-53
Long DDPs, 138
Lotus 1-2-3, 34-35, 212
DAL (Data Access Lan-
guage), 109
LU 6.2 protocol, 274
MacDraw Pro format support,
32-33
Macintosh
AppleTalk transport
protocols
addressing, 137-147
Chooser, 155-165
Phase 1 and 2, 148-155
routers, 166-178
suite of protocols, 127-136
cabling systems
ARCNET, 200
conventional analog
dial-up, 203-204
Ethernet, 187-196
FDDI (Fiber Distributed
Data Interface), 197-199
ISDN (Integrated Ser-
vices Digital Network),
204-205
leased-line services, 205
LocalTalk, 179-187
packet switched, 207
PhoneNET, 182-187
Serial RS-232/422,
199-200
Switched 56K Service,
204
Token Ring, 196-197
wireless, 200-202
X.25 packet switched,
206-207
coax and twinax cards, 276
E-mail, 110-113
Index
formats, 115-116
binary, 121
document interchange
common standcird,
122-123
document interchange
conversion, 124-126
graphics, 118-120
raster, 120-121
word processing, 116-117
integrating
with DEC VAX, 257-267
with IBM mainframes/
midranges, 269-278
with PCs, 21 1-230
with UNIX, 231-248
MacPaint, 1 18
NetPICT diagrams, 371-372,
377, 379, 381
QuickDraw, 118-119
services
Apple Open Collabora-
tion Environment
(AOCE), 105-106
database, 108-110
file sharing, 99-103
mail, 110-113
printing, 103-105
terminal, 106-108
standard networking layers,
31-33
Macintosh Communications
Toolbox (MCT)
AppleTalk/pECnet tool, 264
AppleTalk/ 1 AT tool, 265
CTERM Communications
tool, 262
integrating Macintosh and
UNIX, 241-242
Macintosh File Sharing,
101-102
MaclPX (Novell), 223
MacLAN Connect Gold
(Miramar Systems Inc.), 222
MacLinkPlus (DataViz),
117,214
MacMainFrame (Avatar), 270
MacODA (Apple), 122-123
MacPaint (Macintosh) bitmap
graphics format, 118
MacTCP (Apple), 23-24, 226,
244-247
MacTerminal (Apple), 252
NetPICT diagrams, 377-378
MacWorkStation (United Data
Corporation), 108
MacX (Apple), 25, 240, 252-253
Mail (Microsoft), 212
mail, electronic, see E-mail
Mail for Macintosh
(Microsoft), 111, 256
MailMateQM (Alisa), 112
mainframes (IBM), integrating
with Macintosh
cabling systems
direct connections,
276-277
Ethernet, 278
Token Ring, 278
formats, 273
NetPICT diagrams, 376
services
DAL database, 272
terminal, 269-272
transport protocols, 273-275
Live Wired
Management Information
Base (MIB), 316
mapping networks, 320
Marietta Systems
International, 389
matching
communication layers, 13-15
networking layers, 37-38
Mathematica, 39
Mbps (million bits per
second), 66
MCI, 389
Mail, mail gateways, 1 12
MCT, see Macintosh Commu-
nications Toolbox
MDG Computer Services, 389
MediaWorks’ Executive
Workstation, 271
Medium layer of
communications, 11-15
memory, PRAM (Parameter
RAM), 139
MHS (Novell), mail
gateways, 112
MIB (Management
Information Base), 316
Microsoft Corporation, 389
applications for both Macs
and PCs, 212
Excel, DAL (Data Access
Language), 109
LAN Manager, 35
Mail for Macintosh, 111, 256
Word, data format
translators, 117
million bits per second
(Mbps), 66
Miramar Systems Inc., 389
MacLAN Connect Gold, 222
Mitem Corporation, 389
MitemView, 108
Mitem Vision, 271
modem ports, 26
modulating amplitude, 58
Motif (Open Software
Foundation), 238
Motorola
Altair II wireless networks,
201-202
EMBARC, 201
Computer Group, 389
MPW (Apple), ASCII text, 116
MultiAccess Computing
Corp., 389
multicasting, 84
Phase 2, versus Phase 1
broadcasting, 153-155
multiple
Ethernet networks
FDDI backbone, 309
FDDI backbone WAN,
309-311
routed backbone and
star, 307-308
LocalTalk networks
bridged backbone,
305- 306
Ethernet backbone,
306- 307
routed backbone, 304-305
serially-routed, 302-304
star topologies, single
LocalTalk networks,
300-301
Index
multiprotocol routers, 80-86
with leased-line
services, 205
N
Name Binding Protocol (NBP),
134, 157
names
domain headers (Dls),
177-178
Neuvork Visible Entities
(NVEs), 157-165
zone, AppleTalk, 142-147
Names Information Socket, 160
National Semiconductor
Corporation, 389
NBP (Name Binding Protocol),
134, 157
NCSA Telnet, 235, 241
NE1000/NE2000 (Novell), 228
Neon Software, Inc., 390
LANsurveyor, 320
NetMinder, 322-323
RouterCheck, 321
NetAtlas (Farallon), 320
NetMinder (Neon Software),
322-323
NetMounter (Dayna), 223
NetPICT diagrams, 375
NetPlCT symbols
comparing with OSl
Reference Model, 40
designing networks witli,
285-287
diagrams, 369-382
layers
Cabling, see Cabling
NetPICT layer
Format, see Format
NetPICT layer
matching, 37-38
Transport, see Transport
NetPICT layer
Service, see Service
NetPICT layer
standard Macintosh,
31-33
standard PC, 34-36
swapping, 36-37
NetWare (Novell), 35
integrating Macs and
PCs, 222
NetPICT diagrams, 379-380
SQLNLM, DAL support, 109
Network Filing System (NFS)
(Sun Microsystems), 237-238
gateways with Apple Filing
Protocol (AFP), 87-88
NetPICT diagrams, 379
Network General Corporation,
390
network numbers, 51
AppleTalk, 137-138
Phase 1 and 2 assign-
ments, 149-152
Network OSI layer, 42-43
Macintosh protocols,
134-135
with Macintosh, 45
Network Resources
Corporation, 390
Live Wired
Network Visible Entities
(NVEs), 157-165
Network Vital Signs (Dayna),
331-332
networking cards, see cards
networking layers, see NetPICT
symbols
networks
addressing, 49-50
logical, 50-53
physical, 54-55
analyzers (packet sniffers),
322-323
analyzing traffic levels,
331-332
bandwidth, 59
comparing Ethernet and
LocalTalk, 330-331
broadband, 59
circuit-switched, 47-50
designing, see designing
networks
frames, 56-57
future, 337-339
isolating
high-traffic nodes,
332-335
traffic, 77-78
managing, 313-317
mapping, 320
packet-switched, 47-50
security, 335-336
serially-routed, 76
signaling, 57-63
Nevada Western, 390
Newton (Apple), 29
NFS, see Network Filing
System
NFS/Share (InterCon), 237
nodes, 50
configuring, managing,
313-315
numbers, 50
AppleTalk, 138-142
Novell, Inc., 390
DAL servers, 109
integrating Macs and PCs,
222-224
MHS, mail gateways, 112
NE1000/NE2000, 228
NetWare, see NetWare
NuBus Token Ring card
(Apple), 229
numbers
network, 51
AppleTalk, 137-138,
149-152
node, 50
AppleTalk, 138-142
socket, 51-52
Names Information
Socket, 160
NVEs (Network Visible
Entities), 157-165
0
octets, 19
ON Technology, Inc., 390
Open Document Architecture
(ODA) standard, 122-123
Index
Open Shortest Path First
(OSPF) protocol, 310, 338
Open Software Foundation’s
Motif, 238
Oracle Corporation, 390
databases, accessing.
109-110
OSl Reference Model
layers, 40-43
with Macintosh, 44-45
P
Pacer Softu'are, Inc., 390
AppleTalk implementation,
244
DAL servers, 109
PacerShare, 233, 256
packet-switched networks,
47-50, 206-207
packets
acknowledge control, 139
enquiry control, 139
sniffers (network analyzers),
322-323
transaction release, 134
transaction request, 133
transaction response, 134
PageMaker (Aldus), translating
formats between versions,
121
PAP, see Printer Access
Protocol
Parameter RAM (PRAM), 139
Partner (IPT), 237
passive star topologies. 183-185
single LocalTalk networks,
297-298
Pathway NFS (Wollongong),
237
PATHWORKS for Macintosh
(DEC)
CTERM Communications
Tool, 262
DAL server for VMS, 256
DECnet, 261
E-mail, 256-257
gateways, 88-89
MacTerminal terminal
emulation, 252
MacX X- Window
emulation, 252-253
terminal access, 107
VAXshare, seeVAXshare
PC-NFS (Sun Microsystems),
226
PC/TCP Plus (FTP Softw'are),
226
PCL format (Hewlett-Packard],
35
PCs (personal computers)
integrating with Macs
application-based
services, 211-213
cabling systems, 227-230
formats, 213-218
transport protocols,
218-226
NetPlCT diagrams, 380
standard networking layers,
34-36
Live Wired
PDF (Portable Document
Format), 123
Phase 1 networks
broadcasting, versus Phase
2 multicasting, 153-155
device limits, 148-149
network number
assignments, 149-152
zone names, 143-144
Phase 2 networks
multicasting, versus Phase 1
broadcasting, 153-155
network number
assignments, 149-152
transition routing, 152-153
zone names, 144-147
PhoneNET (Farallon), 182-187
NetPlCT diagrams, 376, 380
PhoneNET PC (Farallon),
219-221
PhoneNET Repeater
(Farallon), 67
PhoneNet Talk PC
(Farallon), 23
Photonics Corporation, 390
infrared devices, 201
physical addressing, 50, 54-55
Physical OSl layer, 41-43
with Macintosh, 45
PICT graphics format, 118-119
files for NetPlCT
symbols, 286
P1CT2 graphics format, 1 18
PICTviewer, 286-287
Planet X (InterCon), 240
Point-to-Point Protocol (PPP),
310, 338
Portable Document Format
(PDF), 123
ports
printer serial, 180
router, 73
serial, 26
PostScript (Adobe), 33, 35, 1 19
Macintosh protocol, 129
PowerPoint (Microsoft), 212
PPP (Point-to-Point Protocol),
310, 338
PRAM (Parameter RAM), 139
Presentation OSI layer, 41-42
Macintosh protocols, 129
with Macintosh, 44
Printer Access Protocol (PAP),
103-105, 131-132
on UNIX computers,
232-233
VAXshare, 253-256
printer serial ports, 26, 180
printers
LaserWriter, 104
NetPlCT diagrams, 370
virtual, 233
managing, 315
PROFS (IBM), 271
mail gateways, 112
programs
agent, 314
integrating Macs and PCs,
211-212
responder, 314
Project (Microsoft), 212
Protocol NetPlCT layer, see
Transport NetPlCT layer
protocol transparent, 66
Index
protocols, 11, 23
Address Resolution
Protocol (ARP), 55
Advanced Peer-to-Peer
Networking (APPN), 274
Advanced Program-to-
Program Communications
(APPC), 274
Apple Data Stream Protocol
(ADSP), 132
Apple Filing Protocol (AFP),
see Apple Filing Protocol
Apple Management
Protocol (AMP), 315
Apple Secure Data Stream
Protocol (ASDSP), 106,
132, 323
Apple Transaction Protocol
(ATP), 133-134
Apple Update Routing
Protocol (AURP), 133,
171-174
AppleTalk Address
Resolution Protocol
(AARP),55, 141-142
AppleTalk Echo Protocol
(AEP), 134
AppleTalk for VMS, 260
AppleTalk Remote Access
Protocol (ARAP), 27-28
AppleTalk Session Protocol
(ASP), 131
AppleTalk suite of, 127-136
Asynchronous, NetPlCT
diagrams, 378
CTERM, 262, 377
Datagram Deliveiy Protocol
(DDP), 134-135, 138
EtherTalk Link Access
Protocol (ELAP), 136, 190
File Transfer Protocol (FTP),
integrating Macintosh and
UNIX, 235-236
Local Area Transport (LAT),
see Local Area Transport
LocalTalk Link Access
Protocol (LLAP), 136
LU 6.2, 274
Name Binding Protocol
(NBP), 134, 157
OSPF (Open Shortest Path
First), 310, 338
Printer Access Protocol
(PAP), see Printer Access
Protocol
Point-to-Point Protocol
(PPP),310, 338
PostScript, 129
QuickDraw, 129
Routing Table Maintenance
Protocol (RTMP), 132-133,
167-168, 171-172
Simple Mail Transfer
Protocol (SMTP), 111
Simple Network
Management Protocol
(SNMP), 315-317
System Application
Architecture (SAA), 274
System Network Architec-
ture (SNA), 274
TCP/IP (Transmission
Control Protocol/ Internet
Protocol), seeTCP/IP
TokenTalk Link Access
Protocol (TIAP), 136
Live Wired
tunneling (encapsulating),
174-178
Zone Information Protocol
(ZIP), 130-131
Q
QuickDraw (Macintosh), 32-33
Macintosh protocol, 129
PICT format, 118-119
QuickMail (CE Software),
111-112
QuickTime for Windows
(Apple), 212
R
Racal-Interlan, Inc., 391
racks for electronic
equipment, 90
radial topologies, see star
topologies
RAM, Parameter (PRAM), 139
ranges, startup, 149
raster formats, 120-121
Rdb databases (DEC),
accessing, 109
relational databases,
accessing, 108-110
repeaters, 66-67
Ethernet, 68
LocalTalk, 67-68
NetPICT diagrams, 376
responder programs, 314
Rich Text Format (RTF), 214
ring topologies, 95
RJ-11 connectors, 182
RouterCheck (Neon
Software), 321
routers, 51, 73-74
AppleTalk, 75-79, 166
updating routing tables,
166-174
backbone
multiple Ethernet
networks, 307-308
multiple LocalTalk
networks, 304-305
exterior, 176
interior, 176
isolating high-traffic nodes,
332-335
multiprotocol, 80-86
with leased-line
services, 205
NetPICT diagrams, 372-375,
378-379
seed, 304-305
serial, multiple LocalTalk
networks, 302-304
troubleshooting, 321
tunneling protocols, 174-178
routing
Bellman-Ford, 169-170
transition, AppleTalk Phase
2, 152-153
vector-distance, 169-170
Routing Table Maintenance
Protocol (RTMP), 132-133,
167-168, 171-172
routing tables, AppleTalk,
132-133
updating, 166-174
Index
RS-422
NetPICT diagrams, 371,
377-378
Serial networks, 199-200
RTF (Rich Text Format), 214
RTMP (Routing Table
Maintenance Protocol),
132-133, 167-168, 171-172
s
SAA (System Application
Architecture), 274
SCSI /Ethernet adapters, 189
SDLC (Synchronous Data Link
Control) connections, 275
security, 335-336
Apple Open Collaboration
Environment (AOCE), 106
seed routers, 304-305
SequeLink (TechGnosis), 109
sequencing, controlling, 131
Serial NB card (Apple), 277
serial ports, 26
printer, 180
Serial RS-232
integrating Macintosh and
DEC VAX, 267
networks, 199-200
Serial RS-422 networks, 199-200
serially- routed topologies, 76
multiple LocalTalk networks,
302-304
servers
Apple Filing Protocol (AFP),
99-103,232-233
Data Access Language
(DAL), 108-110, 234, 272
File Transfer Protocol (FTP),
235-236
managing, 315
NetPICT diagrams, 374
Network Filing System
(NFS)
terminal, 82, 241-242, 262,
269-272
X- Window, 238-241
see also client/server
computing
Service NetPICT layer, 18
designing, 283-285
integrating Macintosh
and IBM mainframes/
midranges, 269-272
and PCs, 211-213
and UNIX, 232-242
standard Macintosh, 32
standard PC, 34
Session OSl layer, 42
Macintosh protocols,
129-132
with Macintosh, 45
Shiva Corporation, 391
FastPath router, 80-81
LanRover, 28, 335
Short DDPs, 138
signaling, 57-63
SimMac (Simware), 272
Simple Mail Transfer Protocol
(SMTP), 111
Simple Network Management
Protocol (SNMP), 315-317
Live Wired
single networks
Ethernet
daisy-chain, 294-296
twisted-pair star, 299-300
LocalTalk
active star, 298-299
bridged star, 302
daisy-chain, 292-294
multiple star, 300-301
passive star, 297-298
Sitka Corporation, 391
SMDS (Switched Multimegabit
Data Service) packet
switched networks, 207
SMTP (Simple Mail Transfer
Protocol), 111
SNA (System Network
Architecture), 274
SNA»ps (Apple)
3270 terminal emulator, 270
gateway, 274, 277
NetPICT diagrams, 381
SNMP (Simple Network
Management Protocol),
315-317
socket numbers, 51-52
Names Information
Socket, 160
Soft-Switch Inc., 391
SoftWriters, Inc., 391
Sonic Systems, Inc., 391
SuperBridge bridge, 187
Source Routing Bridges
(SRBs), 334
SPARCstations (Sun)
AFP/ PAP services, 232-233
DAL servers, 109, 234
integrating Macintosh
and UNIX with Partner
(IPT), 237
speed, electromagnetic
signals, 58
spooling print jobs, 104-105
SprintTALK bridge (Digital
Products), 187
SRBs (Source Routing
Bridges), 334
Standard Microsystems
Corporation, 391
standards
ASCII, 20-21
ANSI X3T9.5; FDDI, 290
IEEE 802.3 (lOBase-F), 289
EIA/TIA 568 andTSB-36,
289
Open Document Architec-
ture (ODA), 122-123
Unicode, 20
wiring, anticipating, 194
X-Window
integrating Macintosh
and UNIX, 238-241
MacX, 252-253
star (radial) topologies, 93-95
active, single LocalTalk
networks, 298-299
bridged, single LocalTalk
networks, 302
multiple
Ethernet networks,
307-308
single LocalTalk networks,
300-301
Index
passive, single LocalTalk
networks, 297-298
single twisted-pair Ethernet
netw'orks, 299-300
with PhoneNET, 183-184
StarController repeater
(Farallon), 67, 94, 183, 185
StarNine Technologies,
Inc., 391
startup ranges, 149
structured wiring implemen-
tations, 311-312
substituting networking
layers, 36-37
suite of protocols, AppleTalk,
127
Data Link OSI layer, 135-136
Network OSI layer, 134-135
Presentation OSI layer, 129
Session OSI layer, 129-132
Transport OSI layer, 132-134
Sun Microsystems
Network Filing System
(NFS), 237-238
PC-NFS, 226
SPARCstations
AFP/PAP services,
232-233
DAL servers, 109, 234
integrating Macintosh
and UNIX, 237
workstations, NetPICT
diagrams, 382
SuperBridge bridge (Sonic
Systems), 187
Switched 56K Service
networks, 204
Switched Multimegabit Data
Service (SMDS) packet
switched networks, 207
Sybase databases,
accessing, 109
Symantec’s Lightspeeed C,
ASCII text, 116
Synchronous Data
Link Control (SDLC)
connecrions, 275
SyncRouter (Engage
Communications, Inc.)
LTNT router, 204
NetPICT diagrams, 375
Synergy Software, 391
System 7, Macintosh File
Sharing, 101-102
System Application
Architecture (SAA), 274
System Network Architecture
(SNA), 274
T
T1 WAN links, 66
tables
Address Mapping Table
(AMT), 55, 141
AppleTalk routing, 132-133,
166-174
Zone Information Table
(ZIT), 130-131
Tagged Image File Format
(TIFF), 120-121
Talaris Systems, Inc., 391
Tandem, DAL servers, 109
Live Wired
TCP/IP (Transmission Control
Protocol/Internet Protocol)
Address Resolution
Protocol (ARP), 55
integrating Macintosh
and IBM mainframes/
midranges, 275
and PCs, 226
NetPICT diagrams, 381
subdividing networks, 81-82
on Macs, 244-247
TechGnosis Inc., 392
SequeLink, 109
TechWorks, 392
tee connectors, 193
Tektronix graphics terminals,
emulating, 106
Telnet terminal service, 241
NetPICT diagrams, 381
Teradata's DBC/1012
databases, accessing, 109
terminal
emulators, 106
3270, 269-270
5250, 270
Advanced Software
Concepts (ASC), 241-242
MacTerminal, 252
NetPICT diagrams,
377-378, 381
servers, 82, 262
services, 106-108
integrating IBM main-
frames/midranges with
Macintosh, 269-272
integrating Macintosh
and UNIX, 241-242
MacTerminal, 252
Terranetics, 392
testing network systems,
318-319
text files for NetPICT
symbols, 286
thickwire (lOBase-5) Ethernet,
93, 190-192, 196
thinwire (lOBase-2) Ethernet,
92, 192-193, 195
Thomas-Conrad Corporation,
392
Thursby Software Systems,
Inc., 392
TIFF (Tagged Image File
Format) format, 120-121
TLAP (TokenTalk Link Access
Protocol), 136
Token Ring, 196-197
AppleTalk support for, 27
integrating Macintosh
and IBM mainframes/
midranges, 278
and PCs, 228-229
and UNIX, 248
isolating high-traffic
nodes, 334
NetPICT diagrams, 370,
372-373, 380
network signals, 61-62
networking cards, 36
ring topologies, 95
zone names, 144-147
TokenPeek (AG Group),
322-323
tokens, 61-62
TokenTalk (Apple), 27
TokenTalk Link Access
Protocol (TLAP), 136
Index
topologies, 91
backbone (bus), 76-77, 92-93
bridged, multiple
LocalTalk nettvorks,
305- 306
Ethernet, multiple
LocalTalk networks,
306- 307
FDDl, multiple Ethernet
networks, 309-311
routed, multiple Ethernet
networks, 307-308
routed, multiple
LocalTalk networks,
304-305
with Ethernet, 196
with PhoneNET, 185
composite, 95
daisy-chain, 91-92
single Ethernet networks,
294-296
single l^calTalk networks,
292-294
ring, 95
serially routed, 76
multiple LocalTalk
networks, 302-304
star (radial), 93-95
active, single LocalTalk
networks, 298-299
bridged, single LocalTalk
networks, 302
multiple Ethernet
networks, 307-308
multiple, single
LocalTalk networks,
300-301
passive, single LocalTalk
networks, 297-298
single twisted-pair
Ethernet networks,
299-300
with PhoneNET, 183-184
traffic
analyzing levels, 331-332
isolating high-talking
nodes, 332-335
transaction release
packets, 134
transaction request
packets, 133
transaction response
packets, 134
transceivers, 93, 191
transition routing, AppleTalk
Phase 2, 152-153
Transmission Control
Protocol/Internet Protocol,
see TCP/IP
transmission errors, cyclic
redundancy checks, 57
Transport NetPlCT layer, 22-25
AppleTalk for UNIX, 244
AppleTalk for VMS, 260
AppleTalk/DECnet
Gateway, 263-264
AppleTalk/LAT Gateway,
264-265
DECnet, 261-262
designing, 284-285
integrating Macintosh
and DEC VAX, 262-263
and IBM mainframes/
midranges, 273-275
and PCs, 218-226
Live Wired
standard Macintosh, 33
standard PC, 35
Transport layer of
communications, 1 1
matching senders and
receivers, 13-15
Transport OSI layer, 42-43
Macintosh protocols,
132-134
with Macintosh, 45
Tribe Computer Works, 392
LocalSwitch bridge, 71, 302
NetPICT diagrams, 376
TribeStar bridge, 305, 332
Trik, Inc., 392
troubleshooting
Chooser, 324-330
finding printers, 165
mapping networks, 320
network analyzers (packet
sniffers), 322-323
routers, 321
testing systems, 318-319
tunneling, 85-86, 174-178
TurboStar repeater (Focus),
68, 94, 185
twinaxial (twinax) cable, 276
twisted-pair (lOBase-T)
Ethernet, 68, 95, 193-194, 196
single star networks, 299-300
type code, 57
u-v
Ultrix (DEC)
AFP/ PAP services, 233
DAL servers, 109, 234
Ungermann-Bass, Inc., 392
Unicode standard, 20
United Data Corporation, 393
MacWorkStation, 108
UNIX
DAL servers, 109
integrating with Macintosh
cabling systems, 248
formats, 242-243
services, 231-242
terminal protocols,
244-247
Simple Mail Transfer
Protocol (SMTP), 111
Unshielded Twisted-Pair
(UTP) wiring, 289
upgrading network software,
314
US Sprint Communications,
393
vampire taps, 191
VAX/VMS (DEC)
Apple Filing Protocol (AFP)
servers, 102-103
DAL servers, 109
integrating with Macintosh
cabling systems, 266-267
formats, 257-259
services, 251-257
transport protocols,
260-265
VAXshare, 102-103,253-256
integrating files between
Macintosh and VAX, 258
PAP spooler, 105
vector-distance routing,
169-170
VersaTerm-Pro, 235, 241
Index
VINES (Banyan), 35, 224
NetPlCT diagrams, 374
virtual LaserWriters, 233
VMS Mail, mail gatev\'ays, 112
VT-series terminals (DEC),
emulating, 106
w
WANs, see Wide Area Networks
WatchTower (InterCon), 317
Webster Computer
Corporation, 393
Wellfleet routers, 85
multiprotocol, 205
White Pine Software, 393
eXodus, 240
Wide Area Networks (WANs),
65-66
conventional analog
dial-up, 203-204
ISDN (Integrated Services
Digital Network), 204-205
leased-line services, 205
multiple Ethernet FDDI
backbone networks,
309-311
NetPlCT diagrams, 374
packet switched, 207
Switched 56K Service, 204
X.25 packet switched,
206-207
wildcard characters, with
Network Visible Entities
(NVEs), 158
WilTel, 393
wireless networks, 200-202
wiring closets, 290-291
The Wollongong Group,
Inc., 392
NFS, 226
Pathway NFS, 237
Word (Microsoft), 212
data formal translators, 117
word processing formats,
116-117
integrating Macs and PCs,
214-215
WordPerfect Corporation, 393
WordPerfect, 212
x-z
X-Window standard
integrating Macintosh and
UNIX, 238-241
MacX, 252-253
X.25 packet switched networks,
206-207
Xinet, 393
K-AShare, 232
K-Spool, 232-233
K-Talk, 244
NetPlCT diagrams, 382
XTND (Claris), data format
translators, 117, 124-125
Zone Information Protocol
(ZIP), 130-131
Zone Information Table (ZIT),
130-131
zones
Default, 145
names, 78-79
AppleTalk, 142-147
KEEP UP TO DATE ON
MACINTOSH CONNECTIVITY
THROUGHOUT THE YEAR.
SUBSCRIBE TO
^Connections
^ T ^ THE TECHNICAL JOURNAL OF MACINTOSH CONNECTIVITY
I f youVe a Macintosh networking and communications
professional you need to stay abreast of this fast-moving
industry and you should be reading Connections: The
Technical Journal of Macintosh Connectivity. Connections is
the only publication devoted exclusively to issues that are
important to people who design and implement Mac
connectivity solutions. Whether you're interested in Mac to
Mac, Mac to PC, Mac to UNIX, or Mac to mainframe
connecti\ity. Connections has valuable information you
cannot afford to be without.
Connections is different because of its focus. We don’t spend
time on video cards, monitors or spreadsheets. We talk
about networking. This focus allows us to go beyond the
topics found in the monthly and weekly magazines.
Connections takes you inside the cutting-edge technology of
such issues as Apple’s Open Collaboration Environment,
Wide Area Networking strategies, and Enterprise Network
Management techniques. Each issue of Connections
provides you with approximately thirty-two pages of in-
depth tutorials and analysis about the issues you face, and
will be facing, on a daily basis. Connections does not accept
any advertising, so ever>^ page contains information you
can use. Information this focused on Macintosh
networking is not available anwhere else.
Connections costs just $195 a year for eight information-
packed issues— a small price to pay for staying on top of the
fastest-moving section of the Mac indiistr>\
You can try Connections: The Technical Journal of Macintosh
Connectivity risk-free. Simply return tlie card below and
you’ll receive your first issue absolutely free. If you decide
Connections isn’t an outstanding resource, just write “cancel”
on the accompanying invoice and we’ll stop the subscription
immediately. Also, you can cancel your subscription at any
time and we will refund the subscription fee for any
remaining issues.
Don’t miss another issue. Start your no-risk subscription
today! Mail or fax a copy of the card below to:
Winehouse Computer Company
20 North Santa Cruz Avenue Tel: 408-354-2500
Los Gatos, CA 95030 Fax: 408-354-2571
I 1
I YES! Please enter my no-risk subscription and send me o free issue of Connections. If I like what I see, HI pay SI 95 and get I
I seven more issues for a total of eight. If not, I'll return the bill marked "cancel", keep the free issue, and owe nothing. I
I □ Payment Enclosed □ US/Canada Subscription: $1 95 Name -
I □ Bill Me □ Overseas Subscription: $250 Company |
■ Charge my credit card: QVisa □ MasterCard Street ■
I Card Number Gty/State/ZIP ■
I Expires Phone Number |
J Signature E-Mail Address j
I I
Making Sense of AppleTalk Networking
Since 1986 !
When it comes to AppleTalk networking, few companies are as qualified as Computer
Methods Corporation. Over the years, our organization has helped hundreds of companies
of various types and sizes make sense of their AppleTalk networking needs. Our staff of
highly trained AppleTalk experts can help you find answers to the tough questions like:
"Hoiu many AppleTalk zone names can my internetwork handle?''
"How can I connect my Macintosh users at remote locations?"
"What is the difference between seeding and non-seeding routers?"
"When should I use a bridge and when should I use a router?"
"What does the term tunneling mean and what are its implications?"
"What are my options for wide-area AppleTalk networks?"
"When will 1 exceed the limitatiorjs of my LocalTalk LANs?"
"Should 1 replace them with Ethernet, Token Ring or FDDI LANs?"
f
T
If AppleTalk networking has you confused, or if you’re interested in
learning more about our full range of consulting services, call us today at...
^ 800 - 969-4360 ^
• AppleTalk Network Integration Experts • Multivendor Systems Integrators
• Certified Apple Developers • Members of the Apple Consultant Relations
• Authorized Macintosh Resellers Team • Apple Training Alliance Partners
COMPUTER METHODS CORPORATION
525 Route 73 South, Suite 300, Marlton, NJ 08053
How to Use
the Disk
his book comes with a disk that contains an
entire library of the NetPICT symbols used
throughout tlie book. For each NetPICT symbol
there are two files: a PICT file that contains the
graphical image, and a text file that contains a
description of the image. The text file has the
same name as the PICT, but with the addition of a
“.txt” filename extension. For example, the PICT file for Apple’s
Internet Router is named Apple Internet Router, while the text
file is named Apple Internet Router.txt.
The disk also contains a viewer application, Ccilled PICTviewer.
PICT viewer is used to view both the NetPICT symbols and the
associated text. PICTviewer provides an easy means for you to view
and read informative data on hundreds of products. Of course,
since the symbols are in the PICT format, they can be easily
incorporated into documents made by your favorite drawing
program. The same can be said for the text (.txt) files. These text
files can be added to your drawings to annotate the figures. Many
drawing programs, such as MacDraw Pro, can import text files
directly. If your drawing program can’t directly import text files,
then you’ll have to copy and paste the text using a text editor or
word processor. By using a drawing program, you’ll be able to
import those NetPICT symbols that represent the fundamental
parts of your present or planned network. You’ll be able to see
where bridges, routers, gateways, or even format translators are
N
required.
The NetPICT files are compressed to save spaa'l You will have to
decompress them and save them to your hard disk in order to use
them. (They will take about 3M of disk space when you have
extracted them.) Double-click on the NetPiCTs.sea icon. You will
be prompted to choose a location to save the extracted files. Wlien
you have done so, click the Save button. The files will self-extract.
To use the NetPICT viewer, you’ll need HyperCard 2.1 (or the
HyperCard Viewer). You might also want to copy PICTviewer to
your hard disk.
1 . Launch the NetPICT viewer by double-clicking its icon.
2. Click three times to close the splash screens.
3. Use the pop-up folder menu to browse through the NetPICT
symbol folder hierarchy.
4. Once you’ve found the appropriate symbol, chck once on the
name that appears in the window directly beneath the pop-
up menu.
The chosen NetPICT symbol will appear in a separate win-
dow to the right of the viewer screen. You can reposition this
window at any time. You can even enlarge it to the size of
your monitor by clicking on the Zoom box located at the
upper right hand corner of the symbol window.
5. To restore the symbol window to the right side of the viewer
window, click once on the magnet icon. This will automati-
cally dock the symbol window to its default position at the
right of the viewer window.
/
A Guide to Net iv or king Macs
LIVE
WIRED
Live Wired makes building networks as
[clear and easy as the Mac itself.
^H<
In this book, Mr. Anders introduces an
^■elementary symbol language that unifies
^Hand simplifies Macintosh network design
vRand system integration. These symbols
I -Uhelp make complex issues simple to
grasp and easy to apply. Each hardware
or software element relevant to net>
working is explained and diagrammed —
so you’re comfortable with their func-
^^tions in the big picture. The result —
properly designed, efficient networks.
LEARN:
• How communication takes place
• Necessary networking
components
• How to design your own Mac
network
Easy network diagramming with
NetPICTs
* How to network Macs and PCs
about the author
Jim Anders
I Live Wired shows you more than just
Mac-to-Mac networking. The book
explores cross-platform networking
.strategies with DOS, UNIX, DEC VAX,
land various IBM platforms. Whatever
you use, however you work with it, you
can design your network with confidence
using this popular method.
He is a Senior Consulting Engineer with
Computer Methods Corporation, specializing
in Macintosh/VAX integration and
CAD/CAM technologies. His reviews and arti-
cles appear regularly in MacUser magazine.
I
Disk includes the popular
[NetPICT symbol encyclopedia
Price: $29.95 US/$37.95 CAN
ISBN 1-56830-015-8
90000
H^en
Books
781568
0015 :
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11