ATM • I Tame Relay • SMDS • Fibre Channel • Iasi Filierne
`l.i:i: l.W-i (l ire Wire) • Cigabil Fibernel • 1111*1*1 • FAN Switching
`ion a ¥ol ilib tfilO ^iliitiU o jo l tfilftl'in y k 'fv ;
`
`m
`
`f 'BI! Ill ?lll ; ill r- H
`HSf Sf- Slii;
`I B O H P ™
`
`M i l s ®
`
`By Dan Blachai
`
`HPE 1037-0001
`
`HPE Co. v. ChriMar Sys., Inc.
`IPR Pet. - U.S. Patent No. 8,902,760
`
`

`

`A Serious Guide to High
`
`• D iscover w hich high-speed data transm ission system designs w ill best m eet your needs
`• Identify key im plem entation and com patibility issues
`• D evelop strategies for im plem enting new netw orking architectures
`• Integrate new technologies w ith existing netw orks
`• M aster key interconnectivity m ethodologies
`
`Maximum Bandwidth is your expert guide to the often com plicated, challenging w orld
`o f high-speed netw orking. W ith in this book you’ll find all the expert advice you need
`to critically evaluate and im plem ent high-speed netw orking technologies.
`
`Learn to identify com patibility issues w ith your current netw ork and cutting-edge
`netw orking technologies. Expert planning advice will guide you through the process
`o f netw ork technology im plem entation an d show you how to successfully integrate
`ATM , Fram e Relay, SM D S, Fibre C hannel, Fast E thernet, Gigabit Ethel. . >
`in* \Vu< ,
`H IP P I, and LAN sw itching technologies w ithin your current netw ork. Ph-’
`.td d " um.il
`chapters detail the m ost efficient w ay to connect your servers arul intr.ua is u> dn
`In tern et and the W orld W ide W eb.
`
`3594
`
`HPE 1037-0002
`
`

`

`Maximum
`Bandwidth
`
`A Serious Gui
`High-Speed Net!
`
`m
`
`p paourf ^ J2 IIIIIIIIIIIIIIIII
`
`T H R IF T B O O K S
`
`G ) 04 16913 913
`
`HPE 1037-0003
`
`

`

`HPE 1037-0004
`
`HPE 1037-0004
`
`

`

`M aximum
`Bandwidth
`
`cjue
`
`HPE 1037-0005
`
`

`

`
`
` MAXIMUM
`BANDWIDTH
`
`Q QUE®
`
`Bandwidth
`
`cjue
`
`HPE 1037-0006
`
`

`

`M aximum
`Bandwidth
`
`Written by Dan Blacharski
`
`puea
`
`HPE 1037-0007
`
`

`

`Maximum Bandwidth
`Copyright© 1997 by Que® Corporation.
`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 quota­
`tions 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 Que Corporation, 201W. 103rd Street, Indianapolis, IN
`46290. You may reach Que’s direct sales line by calling 1-800-428-
`5331.
`Library of Congress Catalog No.: 97-67037
`ISBN: 0-7897-1294-6
`This book is sold as is, without warranty of any kind, either express
`or implied, respecting the contents of this book, including but not
`limited to implied warranties for the book’s quality, performance,
`merchantability, or fitness for any particular purpose. Neither Que
`Corporation nor its dealers or distributors shall be liable to the
`purchaser or any other person or entity with respect to any liability,
`loss, or damage caused or alleged to have been caused directly or
`indirectly by this book.
`99 98 97
`6 5 4 3 2 1
`Interpretation of the printing code: the rightmost double-digit num­
`ber is the year of the book’s printing; the rightmost single-digit
`number, the number of the book’s printing. For example, a printing
`code of 97-1 shows that the first printing of the book occurred in
`1997.
`All terms mentioned in this book that are known to be trademarks
`or service marks have been appropriately capitalized. Que cannot
`attest to the accuracy of this information. Use of a term in this book
`should not be regarded as affecting the validity of any trademark or
`service mark.
`Screen reproductions in this book were created using Collage Plus
`from Inner Media, Inc., Hollis, NH.
`
`HPE 1037-0008
`
`

`

`Contents at a Glance
`
`| Maximum Bandwidth______________________________
`
`1 The Evolution of High-speed Networking 5
`2 Asynchronous Transfer Mode 13
`3 Frame Relay 51
`4 Fast Ethernet 77
`5 Gigabit Ethernet 105
`6 IsoEthernet 137
`7 Switched Multimegabit Data Service (SMDS) 159
`8 Synchronous Optical Network 185
`9 HIPPI 211
`10 Fibre Channel 227
`11
`IEEE 1394 (FireWire) 249
`12 High-Speed Telephony and Internet Access 261
`
`Appendixes
`A Glossary 275
`B A Brief History of Networking 307
`C OUI Listing 315
`
`Index 349
`
`HPE 1037-0009
`
`

`

`HPE1037-0010
`
`HPE 1037-0010
`
`

`

`Table of Contents
`Introduction
`1
`What’s In Store for the Future? 2
`Representations Used in this Book 3
`
`Maximum Bandwidth
`
`1 The Evolution of High-Speed Networking 5
`Faster LANs 6
`Analyzing the Need for a High-Speed Networking Architecture 7
`Addressing Bottleneck Issues Without High-Speed Networking 9
`Deciding on a High-Speed Networking Architecture 10
`Cost Analysis 11
`Integration with Existing Infrastructure 12
`Summary 12
`
`13
`
`2 Asynchronous Transfer Mode (ATM)
`Overview of ATM 14
`ATM Background 14
`Applications for ATM 15
`ATM Speed 18
`ATM Bandwidth Utilization 19
`Integrating ATM Into Existing Networks 20
`ATM Cost Savings 20
`ATM Security 21
`Standards Development (ATM Forum) 21
`ATM Implementation and Infrastructure 23
`Managing ATM 26
`LAN Emulation (LANE) 32
`Virtual LANs (VLANs) 35
`Integrating Legacy LANs with ATM 36
`Integrating IP with ATM 40
`MPOA (Multiprotocol over ATM) 41
`
`HPE 1037-0011
`
`

`

`viii
`
`| Maximum Bandwidth
`
`Product Overview 43
`Fore Systems 43
`Crosscom 45
`Cisco Systems 46
`Cascade Communications Corporation 47
`End2End Network Solutions 49
`Summary 49
`
`3 Frame Relay 51
`Overview of Frame Relay 52
`Frame Relay Background 52
`Applications for Frame Relay 53
`Types of Frame Relay Networks 54
`Frame Relay Speed 56
`Standards Development (Frame Relay Forum) 56
`Implementation and Infrastructure 57
`Routing Frame Relay 58
`Frame Relay Media and Topology 59
`Vendor-Specific Frame Relay Implementations 60
`Definitions 60
`Frame Relay and ATM 62
`Frame Relay Interworking 62
`Frame User Network Interface (FUNI) 64
`Voice and Fax over Frame Relay 64
`SNA and Frame Relay 69
`Pricing Structures 72
`Product Overview 73
`MCI Communications Corporation 73
`ACT Networks, Incorporated 74
`Netrix 75
`Cisco Systems 75
`Summary 76
`
`4 Fast Ethernet 77
`Overview of Fast Ethernet 78
`Evolution of Fast Ethernet 78
`
`HPE 1037-0012
`
`

`

`Contents
`
`|
`
`lx
`
`Standards Development (IEEE) 79
`Implementation and Infrastructure 80
`CSMA/CD 81
`Topology 81
`Cabling 81
`The Layers of Fast Ethernet 83
`Hub Types 85
`Bridging and Routing 90
`Migration to Fast Ethernet 90
`Network Management 92
`Virtual LANs 94
`Ethernet Switching 95
`Coexistence with Other Network Technologies 96
`Fast Ethernet Basics 96
`lOOVG-AnyLAN 97
`Troubleshooting Fast Ethernet Components 99
`Product Overview 100
`Farallon Computer 100
`Cogent Data Technologies 101
`Asante 101
`3Com 101
`Cisco Systems 102
`Cnet 102
`Summary 103
`
`105
`5 Gigabit Ethernet
`Overview of Gigabit Ethernet 107
`Service Guarantees 108
`Virtual LANs 108
`Standards Development 110
`Gigabit Ethernet Goals 111
`Resource Reservation Protocol (RSVP) 111
`Fibre Channel Standards 112
`Implementation and Infrastructure 114
`Performance Issues 114
`Cabling 118
`
`HPE 1037-0013
`
`

`

`Upgrading to Gigabit Ethernet 120
`Gigabit Medium Independent Interface 128
`FDDI and Gigabit Ethernet 129
`ATM and Gigabit Ethernet 130
`Fibre Channel and Gigabit Ethernet 131
`Product Overview 132
`Cisco Systems 132
`Packet Engines, Inc. 132
`GigaLabs 133
`Alteon Networks, Inc. 133
`NBase Communications 134
`XLNT 134
`Summary 135
`
`6
`
`137
`tsoEthernet
`Overview of IsoEthernet 138
`Real-Time Data over 10Mbps Ethernet 138
`No Major Upgrade Required 139
`Standards Development 139
`Isochronous Network Communication Alliance 140
`IEEE Standard 140
`Implementation and Infrastructure 141
`Combining Ethernet with ISDN 142
`Modes of Operation 145
`Multimedia Over IsoEthernet 145
`IsoEthernet and ATM 148
`Quality of Service Guarantees 148
`Topology 152
`Advantages of IsoEthernet 152
`Product Overview 153
`Ascom Nexion 153
`National Semiconductor 154
`Incite 154
`Luxcom 155
`Summary 157
`
`HPE 1037-0014
`
`

`

`Contents
`
`| xi
`
`159
`7 Switched Multimegabit Data Service (SMDS)
`Overview of Switched Multimegabit Data Service (SMDS) 161
`Standards Development 164
`Implementation and Infrastructure 164
`Addressing Scheme 165
`Tariffs and Access Classes 166
`Configuration 167
`Connectionless Broadband Data Service 170
`SMDS Internetworking 171
`Vendors 179
`Bay Networks 180
`Cisco Systems, Inc. 180
`MultiAccess Computing Corporation 180
`QPSX Communications Limited 181
`ADC Kentrox 181
`Digital Link Corporation 182
`Alcatel Network Systems 182
`AT&T Network Systems 182
`Cascade Communications Corp. 183
`Brooktree Corporation 183
`UUNET Technologies 184
`Summary 184
`
`185
`
`8 Synchronous Optical Network (SONET)
`Overview of SONET 186
`SONET and Synchronous Digital Hierarchy (SDH) 187
`Advantages of SONET 188
`Implementation and Infrastructure 188
`A Step Beyond Synchronous Transfer Mode 188
`Self-Healing Mechanism 189
`Carriers 190
`The First SONET Ring 191
`Elements of SONET 192
`Interleaving 197
`
`HPE 1037-0015
`
`

`

`Maximum Bandwidth
`
`Ring Architecture 197
`Virtual Tributary 199
`Jitter and Wander 199
`Standards 200
`SONET Internetworking 201
`High Performance Parallel Interface over SONET 201
`ATM over SONET 202
`Vendors 205
`Hitachi 206
`MetaSolv Software 206
`Northern Telecom 207
`Ameritech 208
`Optivision 208
`Bell Atlantic 208
`Summary 209
`
`211
`9 HiPPI
`Overview of HiPPI 212
`The Evolution of HiPPI 213
`Beyond Supercomputers 214
`Implementation and Infrastructure 214
`HiPPI Protocols 215
`HiPPI 6400 218
`HiPPI Goes to the Movies 219
`Integration with Other Technologies 219
`Standards Development 221
`Serial HiPPI 221
`Product Overview 222
`Essential Communications 222
`Applied Micro Circuits Corporation 222
`GigaLabs 223
`Broadband Communications Products (BCP) 223
`Cray Research (A Silicon Graphics Company) 223
`Hewlett-Packard 223
`IBM Corp. 224
`
`HPE 1037-0016
`
`

`

`Contents
`
`Optivision 224
`Silicon Graphics 224
`Summary 224
`
`227
`10 Fibre Channel
`Overview 228
`Standards Development 229
`Implementation and Infrastructure 230
`Types of Optics Used in Fibre Channel 231
`Fibre Channel Frame Format 232
`Encoding 232
`Protocol Stack 233
`Topology 235
`Cabling 238
`Fibre Channel and Other Network Technologies 240
`Vendors 242
`Emulex Corporation 242
`Adaptec 243
`Ancot Corporation 243
`Digital Equipment Corporation 244
`Fujikura America 244
`Gadzoox Microsystems 244
`Hitachi Data Systems 245
`LSI Logic 245
`Raidtec Corporation 245
`Seagate Technology 246
`Sun Microsystems 246
`W.L. Gore & Associates 246
`Summary 246
`
`11
`
`249
`IEEE 1394 (FireWire)
`Overview of IEEE 1394 250
`Implementation and Infrastructure 250
`Industry Support 252
`Universal Serial Bus 252
`
`HPE 1037-0017
`
`

`

`Maximum Bandwidth
`
`Physical Infrastructure 253
`Digital Video and Multimedia 254
`Advantages 254
`1394 and Asynchronous Transfer Mode 255
`Standards Development 255
`Updates and Improvements 256
`Operating System Support 256
`Vendors 257
`3A International 257
`Adaptec 257
`Kenwood 258
`Matsushita 258
`Sony 258
`Skipstone 258
`Texas Instruments 259
`Toshiba 259
`Philips Semiconductors 259
`Summary 260
`
`12 High-Speed Telephony and Internet Access 261
`ISDN 262
`High-Speed Telephony: xDSL 262
`Asynchronous Digital Subscriber Line (ADSL) 263
`HDSL 266
`VDSL 266
`Standards 266
`DSL and ATM 267
`Vendors 268
`Multilink Channel Aggregation 268
`Intranets 270
`Data Center Intranet 271
`Campus Intranet 272
`Wide Area Intranet 272
`Cable Access 273
`Satellite 274
`Summary 274
`
`HPE 1037-0018
`
`

`

`Contents
`
`I xv
`
`| Appendixes
`
`A Glossary 275
`
`B A Brief History of Networking 307
`The Dividing Line 308
`Early Network 309
`The Internet and More Standardization 311
`Multi-User Operating Systems Advance Networking 312
`The Wave of the Future 313
`
`C OUI Listing 315
`
`Index 349
`
`HPE 1037-0019
`
`

`

`HPE 1037-0020
`
`HPE 1037-0020
`
`

`

`Credits
`
`PRESIDENT
`Roland Elgey
`SENIOR VICE PRESIDENT/PUBLISHING
`Don Fowley
`
`PUBLISHER
`Joseph B. Wikert
`
`PUBLISHING DIRECTOR
`Brad R. Koch
`GENERAL MANAGER
`Joe Muldoon
`
`EDITORIAL SERVICES DIRECTOR
`Elizabeth Keaffaber
`
`MANAGING EDITOR
`Thomas F. Hayes
`DIRECTOR OF MARKETING
`Lynn E. Zingraf
`ACQUISITIONS EDITOR
`Tracy M. Williams
`SENIOR PRODUCT DIRECTOR
`Lisa D. Wagner
`
`PRODUCT DIRECTOR
`Becky Campbell
`PRODUCTION EDITOR
`Julie A. McNamee
`EDITORS
`Lisa M. Gebken
`Brian Sweany
`Tom Lamoureux
`San Dee Phillips
`STRATEGIC MARKETING MANAGER
`Barry Pruett
`WEB MASTER
`Thomas H. Bennett
`Composed in Century Old Style and ITC Franklin Gothic by Que Corporation.
`
`INDEXER
`Chris Barrick
`
`PRODUCT MARKETING MANAGER
`Kristine R. Ankney
`
`ASSISTANT PRODUCT MARKETING
`MANAGER/DESIGN
`Christy M. Miller
`ASSISTANT PRODUCT MARKETING
`MANAGER/SALES
`Karen Hagen
`TECHNICAL EDITOR
`Brad Lindaas
`MEDIA DEVELOPMENT SPECIALIST
`David Garratt
`TECHNICAL SUPPORT SPECIALIST
`Nadeem Muhammed
`
`ACQUISITIONS COORDINATOR
`Tracy M. Williams
`SOFTWARE RELATIONS COORDINATOR
`Susan D. Gallagher
`EDITORIAL ASSISTANT
`Virginia Stoller
`BOOK DESIGNER
`Ruth Harvey
`COVER DESIGNER
`Glenn Larsen
`PRODUCTION TEAM
`Kay Hoskin
`Heather Howell
`Tim Neville
`Sossity Smith
`Lisa Stumpf
`
`HPE 1037-0021
`
`

`

`To my wife Lotus and son Shanti; and to Joe, Kevin, Jeff, and all the rest of
`my old cronies who I haven’t seen in so many years.
`
`HPE 1037-0022
`
`

`

`About the Author
`
`Dan W. Blacharsld is a technology and business writer, novelist, and satirist with several
`years experience. He has written several articles and books, and currently works out of
`his home in Santa Cruz, California.
`
`HPE 1037-0023
`
`

`

`Acknowledgments
`
`I would like to acknowledge and thank Tracy Williams of Que, for her continual support
`and encouragement throughout the entire production of this book. I would also like to
`acknowledge the fine work of development editor Becky Campbell, who was a great help.
`I must also tip my hat to many other individuals involved, from the artists to the typeset­
`ters, and from the technical editors to the salespeople.
`Thanks and appreciation are also due to the many vendors, manufacturers, and
`developers who were kind enough to provide me with information, photographs, and
`product data.
`And lastly, a nod of thanks goes to my son Shanti, for taking me away from my computer
`every now and then and reminding me what a spring day looks like.
`
`HPE 1037-0024
`
`

`

`We’d Like to Hear from You!
`QUE Corporation has a long-standing reputation for high-quality books and products. To
`ensure your continued satisfaction, we also understand the importance of customer ser­
`vice and support.
`
`Tech Support
`If you need assistance with the information in this book or with a CD/disk accompanying
`the book, please access Macmillan Computer Publishing’s online Knowledge Base at:
`http://www. superlibrary, com /general/support
`Our most Frequently Asked Questions are answered there. If you do not find the answer
`to your questions on our Web site, you may contact Macmillan Technical Support by
`phone at 317/581-3833 or via e-mail at support@mcp.com.
`Also, be sure to visit QUE’s Desktop Applications and Operating Systems team Web re­
`source center for all the latest information, enhancements, errata, downloads, and more:
`http ://www. quecorp. com/desktop_os/
`
`Orders, Catalogs, and Customer Service
`To order other QUE or Macmillan Computer Publishing books, catalogs, or products,
`please contact our Customer Service Department:
`Phone: 800/428-5331
`Fax: 800/835-3202
`International Fax: 317/228-4400
`Or, visit our online bookstore:
`http://www.mcp.com/
`
`Comments and Suggestions
`We want you to let us know what you like or dislike most about this book or other QUE
`products. Your comments will help us to continue publishing the best books available on
`computer topics in today’s market.
`
`HPE 1037-0025
`
`

`

`Lisa Wagner
`Senior Product Director
`QUE Corporation
`201 West 103rd Street, 4B
`Indianapolis, Indiana 46290 USA
`Fax: 317/581-4663
`E-mail: lwagner@que.mcp.com
`Please be sure to include the book’s title and author, as well as your name and phone or fax
`number.
`We will carefully review your comments and share them with the author. Please note that
`due to the high volume of mail we receive, we may not be able to reply to every message.
`Thank you for choosing QUE!
`
`HPE 1037-0026
`
`

`

`Introduction
`
`Iri the earliest days of networking, dumb terminals were
`used to send data to a mainframe. This concept was
`decidedly one-sided, and if someone wanted a bit of
`information, it was necessary to approach the high
`priests in the glass house, perhaps bring an offering of
`some sort, and wait two or three weeks. Then, the
`report you got would invariably not be what you had
`originally requested.
`Then, we got PC networks, run by file-sharing network
`operating systems such as NetWare. We still couldn’t
`get at that data behind the glass wall, but we could
`share printers and text files with one another. However,
`these LANs were limited in terms of bandwidth. For
`example, although Ethernet runs at 10Mbps, users
`typically enjoy a maximum of only about 4Mbps, which
`is then further divided by the number of users on the
`network. For example, if there are 10 users on the
`network, each user gets only 400Kbps of bandwidth.
`Hardly enough for multimedia! Although at one time
`this was more than adequate, the increasing demand
`for bigger and more critical applications on the LAN
`has created more traffic jams and congestion, which, in
`turn, drives the need for high-speed networking.
`
`HPE 1037-0027
`
`

`

`2
`
`I
`
`Introduction
`
`Eventually, end users decided they wanted to get at those massive stores of corporate
`data, and wanted to transmit multimedia files as well. However, sending a multi-megabyte
`file full of images overburdened those early PC networks, which were, after all, designed
`to handle data only, and not very much of it, at that.
`Larger files, multimedia, and greater demand for easy access to corporate data all contrib­
`uted to the need for high-speed networking technology, such as Fast Ethernet and ATM.
`Another major factor in the drive to high-speed networks is the move to intranets, a tech­
`nique that uses Internet protocols to send and receive data both internally and externally.
`The open nature of TCP/IP, the dominant Internet/intranet protocol, has created unparal­
`leled levels of data access. Users, regardless of location or platform, can connect to the
`internal network from any dial-up Internet account, through almost any ordinary Web
`browser. Making data access this easy comes with a price: More people will want more
`data. Because it’s so easy to get, those users will probably access it more frequently. For
`years, networking was a frustrating experience for the end user, who faced delays,
`crashes, and dependence on a centralized information system. Thus, as end users wised
`up, took advantage of new technology, and then asked for more, the need for high-speed
`networking arose.
`High-speed networking technologies, such as Fast Ethernet, Frame Relay, FDDI, and
`ATM, are just now beginning to bring massive changes to networking. The desktop PC
`revolution made PCs inexpensive and widely accessible; advances in high-speed network­
`ing will similarly bring access to large databases, graphics, and multimedia to the masses.
`The endless delays, bottlenecks, and the often-heard comment “my, the computer’s slow
`today” may be a thing of the past. These new high-speed technologies are increasingly
`affordable, and many are likely to become commodity items within a few years.
`In the following chapters, we will take a closer look at these high-speed technologies, their
`histories, usage, and implementation. We will see how 10Mbps Ethernet has evolved into
`100Mbps, and then 1,000Mbps Ethernet; and how ultra-high channel technologies are
`being used to connect high-speed devices.
`
`What’s In Store for the Future?
`
`In the future, the “Information Superhighway” will bring high-speed Internet access,
`video-on-demand, and networked appliances to every home. However, this future cannot
`be accomplished with traditional networking technology. An all-optical network would
`remove the bottlenecks common to traditional networks, and thereby potentially increase
`the amount of data carried by each fiber a hundred-fold.
`
`HPE 1037-0028
`
`

`

`Representations Used in this Book
`
`| 3
`
`The first step towards such a network has been taken by the National Transparent Optical
`Network Consortium (NTONC), a group of researchers, manufacturers, and telecommu­
`nications providers led by Lawrence Livermore National Laboratory. Members of the
`consortium include Pacific Bell, Sprint, Northern Telecom, United Technologies Re­
`search, Hughes Aircraft, Rockwell International, and Columbia University. NTONC has
`proposed a $40 million project that would create a prototype optical network in California.
`Eventually, NTONC hopes to develop technology that could be used to carry up to 100
`Terabits per second (Tbps) of data—a capacity far beyond anything currently available or
`underway. At this phenomenal rate, data would be streaking across the network at ten
`million times the rate of standard 10Mbps Ethernet.
`The NTONC project hopes to create a testbed network to demonstrate wavelength divi­
`sion multiplexing (WDM) device technologies and control strategies required to develop
`a terabit-per-second optical network. These speeds will be achieved by using WDM as a
`means of expanding fiber system bandwidth by allowing multiple colors of light to be sent
`over a fiber that currently accommodates only one color. The group will deploy a four-
`node, bi-directional ring network carrying both OC-48 Synchronous Optical Network
`(SONet) and ATM traffic.
`Initially, the project, which will be funded in part by the Defense Advanced Research
`Projects Agency (DARPA), would establish a network around the San Francisco Bay Area,
`using Pacific Bell and Sprint’s existing fiber optic cables. Later, the network would expand
`to Southern California.
`As is the case with new technologies, it is necessary for us to face a slew of acronyms and
`new terms. Even to the most technically proficient among us, these can be confusing. A
`glossary has been added to the end of this book to serve as a quick reference to these
`new terms.
`
`Representations Used in this Book
`
`There are several graphical representations in this book to make it easier to read and
`understand. The following items point you to specific information:
`
`ON THE W EB
`This icon and format signal URL addresses for the Internet and World Wide Web of places that
`have related products or information, such as the Que home page at:
`http:Wwwwni.quecorp.com
`
`HPE 1037-0029
`
`

`

`Introduction
`
`N O T E Notes give advice or general information related to the specific topic.
`
`CAUTION
`This paragraph format warns of hazardous procedures that may cause irreparable damage.
`
`W hat About Sidebars?
`Sidebars are supplementary material that expands on the specific topic being discussed. While
`this information is not vital to the discussion, it may provide valuable insight or further
`explanation.
`
`HPE 1037-0030
`
`

`

`C H A P T E R
`
`The Evolution of High-
`Speed Networking
`
`A Lone time, 10Mbps of bandwidth was all anyone ever
`
`needed. Those days, however, have gone out with 10M
`hard drives and black-and-white screens. Trends in
`both management and technology have driven the
`need for faster networks. Companies are “leaner and
`meaner,” middle management has been eliminated,
`and end users are being given more responsibility.
`Fortunately, the technology has evolved at the same
`time, allowing these end users to gain faster and
`greater access to corporate data of all types. !
`
`HPE 1037-0031
`
`

`

`6
`
`| Chapter 1 The Evolution of High-Speed Networking
`
`Faster LANs
`
`The old 80/20 rule, which states that 80 percent of network traffic exists locally, or within
`the LAN segment, and 20 percent in the backbone, is no longer valid. This design model
`held that an internetwork, which was generally created by connecting LANs with routers,
`ran most traffic over each individual LAN. Routers were optimized to handle this traffic
`pattern. Later, however, companies started to centralize servers on the corporate back­
`bone, creating a new pattern of traffic and destroying the old 80/20 rule. Furthermore,
`increased segmentation in a traditional routed network has led to a much higher percent­
`age of backbone traffic, resulting in more congestion. Multiple segments on a shared
`backbone result in tremendous congestion, as more and more traffic competes for a piece
`of that 10Mbps of bandwidth (see Figure 1.1). In this shared scenario, each segment gets
`only a proportional share of the 10Mbps; that is, if there are six nodes attached to the
`shared hub, each one only gets one-sixth of the 10Mbps.
`
`FIG. 1.1
`A shared backbone
`can result in network
`congestion.
`
`Shared media hub
`
`There are dozens of ways to correct this bottleneck problem, including deployment of a
`switched network. Figure 1.2 shows the same network, divided into segments, with each
`segment receiving its own 10Mbps of bandwidth.
`The switching paradigm can be made even more efficient by adding Asynchronous Trans­
`fer Mode (ATM) technology, which adds a high-speed connection between switches and
`other devices, Quality of Service (QoS) guarantees, and connectivity with legacy LANs.
`Token Ring networks have also suffered from the same bottleneck problems as Ethernet.
`The advent of Ethernet switching has effectively eliminated the technical advantages of
`Token Ring. That is, Token Ring’s deterministic nature had been advantageous because it
`eliminated the possibility of packet collisions. However, this advantage is minimized when
`compared with an Ethernet switching environment, where collisions are far less common
`than in a routed Ethernet environment.
`
`http://www.quecorp.com
`
`HPE 1037-0032
`
`

`

`Analyzing the Need for a High-Speed Networking Architecture
`
`| 7
`
`FIG. 1.2
`A segmented network
`eliminates bottle­
`necks.
`
`Switching hub
`
`Although Token Ring may not have a technical advantage over Ethernet any longer, many
`IBM shops have large Token Ring investments that they may wish to preserve. Short of a
`wholesale migration to ATM, there are some solutions. Although moving to Fiber Distrib­
`uted Data Interface (FDDI) can solve Token Ring bottleneck problems on a short-term
`basis, the problems that result from shared bandwidth still exist. Segmentation of the LAN
`offers a performance boost, although Token Ring switching is still a fairly new, and fairly
`expensive technology. The switching approach merely breaks up the ring into multiple
`smaller rings, which are then linked together with switches. Because there are several
`smaller rings, each user will not have to wait as long for a token to come around.
`Before a Token Ring switching environment is deployed, a serious cost/benefit analysis
`should be undertaken. Management will often find that in the long run, a switched Token
`Ring network will be less advantageous from a cost perspective than moving to a switched
`Ethernet or ATM environment.
`
`Analyzing the Need for a High-Speed
`Networking Architecture
`
`Of course, not everyone needs a high-speed networking architecture. A ten-person com­
`pany running basic word processing and spreadsheet applications would be wasting
`money to deploy such a system. In fact, in many cases, deploying ATM is simply over­
`kill—sort of like using an Indy racecar to drive to the corner supermarket. If you do need
`a high-speed network, it may not be necessary to deploy it throughout the enterprise.
`
`HPE 1037-0033
`
`

`

`8
`
`I Chapter 1 The Evolution of High-Speed Networking
`
`Some workgroups dealing with high-bandwidth applications may require the speed,
`whereas others may not; or it may make sense to deploy a high-speed backbone only.
`If you think you may need a high-speed network, the first thing to do is take a look at the
`existing network infrastructure, see where the bottlenecks are, and determine what your
`current and future needs will be. The best way to approach this is with a formal “Needs
`Analysis.” This process should be implemented by a committee, which includes not only
`the Information Services (IS) organization, but management, and representatives from
`each department in the company. At least some of the departmental representatives
`should be end users; ideally, each department would have two delegates (one manage­
`ment and one end user). This analysis should examine at least the following issues
`in detail:
`II What type of data is currently being sent across the network, and what type of data
`is expected to be sent over the next three years?
`■ What forces are behind the need for a high-speed network? (Bigger files, broader
`access, multimedia, corporate intranet?)
`■I Does every department have a need for high-speed networking, or is it limited to
`certain areas? If so, pinpoint those areas.
`II What problems exist in the current network infrastructure? What are the alterna­
`tives for addressing these problems?
`Another critical component of the initial analysis is a complete inventory and diagram of
`the existing network, including servers, segments, remote sites, and peripheral devices.
`All bottlenecks, suspected bottlenecks, and potential bottlenecks should be pinpointed on
`this diagram. From this existing diagram, and armed with information about bottlenecks,
`you can begin mapping out your future network.
`Make sure to include any new, or potentially new sites that you plan to add, new applica­
`tions, and new or upgraded equipment. When possible, hubs should be replaced with
`switches to maximize bandwidth to each individual end user. Generally, Token Ring
`workgroups should be replaced with switched Ethernet environments, if the investment is
`not too large. And in general, even if you are not planning to deploy a high-speed technol­
`ogy such as ATM immediately, this changeover should be kept open as a possibility, and
`any new equipment acquired should be purchased with high-speed expansion in mind.
`Cabling, for example, is an especially critical consideration when planning for the future.
`Although CAT-3 UTP wiring may be more than adequate for the present, deploying CAT-5
`UTP will leave the door open to an ATM deployment in the future.
`
`http://www.quecorp.com
`
`HPE 1037-0034
`
`

`

`Addressing Bottleneck Issues Without High-Speed Networking
`
`| 9
`
`Addressing Bottleneck Issues Without
`High-Speed Networking
`
`After the initial needs analysis, it should become clear whether or not a high-speed infra­
`structure is needed, or whether any existing bottlenecks could be resolved without such a
`drastic measure. Although for a larger network, migration to a high-speed technology is
`probably the way to go, a smaller network may benefit from interim techniques, such as
`segmentation or performance optimization. Ultimately however, if significant growth is
`foreseen for the future, a migration to high-speed networking is desirable, even if it is
`more than what is needed for the present.
`Performance tuning and optimization can take several different forms. Managers may
`wish to apply one or all of them, although network changes should always be executed
`one at a time in order to get an accurate evaluation of the results of each change. Adding
`network changes in this cautious manner will also greatly assist in troubleshooting when
`the inevitable installation glitch occurs. Following are some simple areas to start optimiz­
`ing your network.
`1. The first place to start is to simply discover whether or not you have the latest
`device drivers, bug fixes, and patches in place.
`2. Network performance can often be enhanced with an upgrade of resources, includ­
`ing cabling, PCs, and other networking devices such as routers and switches.
`Upgrading the network hosts and the links that connect them is certainly a valid
`approach, although often the most costly one.
`3. Tuning, load balancing, and prioritization. In a distributed environment, load
`balancing helps to “even out” the processing, so that one processor is not overbur­
`dened while another is not being fully utilized. Prioritization is becoming an increas­
`ingly common feature of networks; this simply assigns different values to different
`end users or processes. Under a prioritized scheme, a given network service would
`be given to a high-priority user before the low-priority user. When applied to threads
`in a multithreaded environment, such as Windows NT, a thread assigned high
`priority status will get a larger share of processor time when the processor is busy.
`4. Re-visit network design. As the network’s topology evolves, it may be beneficial to
`rearrange the bridges and routers, or apply segmentation and switc