In order to work with today's networks that handle integrated text, data, and interactive voice and video, you must possess a thorough understanding of switches and the newest technologies that make sophisticated network services possible.
Building Switched Networks provides a comprehensive, technical survey of the networking technologies that comprise the core of evolving LAN and WAN infrastructures. This book gives you essential background information, clear descriptions of relevant technologies, and an understanding of how those technologies will be employed throughout networks in the near future. In particular, the text focuses on developments that support our increasing demand for network bandwidth--multilayer switching, delivery guarantees, and multicasting--and examines performance issues, resource allocation, network policy, and network services.
Using a ground-up approach, the book begins with network demands, examines various kinds of available switches, and progresses to the state-of-the-art technologies that are quickly permeating our networking infrastructure. Specifically, you will learn about:
Building Switched Networks then widens the scope of discussion beyond the details of these technologies to examine the demands that will be placed on networks in the future and successful management strategies for meeting those demands.
(Each chapter begins with "Introduction" and concludes with "Conclusions".)
Organization of This Book.
1. Our Demands for Networking.
Key Problems Addressed.
Examples of How We Use Networking Today.
Tomorrow's Networking Requirements.
The Main Goal of Networking.
Incorporating Features of Phone Networks.
The Main Demands on Networking.
Critical Components of Tomorrow.
Behind the Scene Network Needs.
Key Solutions Offered.
The Prerequisite Information Challenge.
The Chapter Two Road Map.
The ISO OSI Reference Model.
Layer 1: Physical.
Layer 2: Data Link.
Layer 3: Network.
Layer 4: Transport.
Layers 1, 2, and 3.
The Big Picture.
Twisted Pair and Fiber.
Connectionless versus Connection-Oriented.
ATM and ATM LANE.
Physical Topology in the LAN.
Physical Topology in the WAN.
Point-to-Point Connections and Cloud Technologies.
Important TCP/IP Concepts.
What is DHCP.
Distance-Vector and Link-State Protocols.
Routing in the LAN.
Routing in the WAN.
WWW and HTTP.
Key Solutions Offered.
Switching Comes from the Telephone Network.
Switches Are "Designed" for the LAN or WAN.
Collision Domains and Broadcast Domains.
Types of LAN Switches.
Types of WAN Switches.
Packet and Cell Switching Technologies.
Frame Relay Switching.
FDDI and Token Ring Switching.
Ethernet, Fast Ethernet, and Gigabit Switching.
Gigabit Ethernet and ATM.
What is a VLAN.
VLAN Creation Techniques.
What is Tagging?
Tag Header Organization.
Key Solutions Offered.
Routers Can Be Traffic Bottlenecks.
Multilayer Switching in a Nutshell.
Fast IP and NHRP.
Multiprotocol over ATM (MPOA).
Routing Combined with Connection-Oriented Services.
The IETF MPLS Effort.
MPLS Functional Requirements.
Important MPLS Terminology.
Key Features of MPLS.
How MPLS Works.
Label Assignment: Local versus Egress Control.
Cell Switched Routers (CSR).
Key Solutions Offered.
When Does Delay Matter.
What Causes Delay, Anyway.
Overprovision, Precedence, Dedicated Resource.
The Guaranteed Reservation Paradox.
Quality of Service (QoS) Basics.
What is QoS, Anyway?
Bandwidth Management and Queues.
Traffic Management Control.
Random Early Discard (RED).
The ATM World.
ATM's Built-In QoS.
ATM Service Classes.
The IP World.
Resource ReSerVation Protocol (RSVP).
Integrating ATM with RSVP.
Type of Service (TOS), OSPF, QOSPF, and QoS.
Key Solutions Offered.
IP Multicast in a Nutshell.
Internet Group Management Protocol (IGMP).
Distance-Vector Multicast Routing Protocol (DVMRP).
Multicast Open Shortest Path First (MOSPF).
Core-Based Trees (CBT).
Protocol Independent Multicast.
Dense Mode (PIM-DM).
Sparse Mode (PIM-SM).
Multicast over ATM (MARS).
Multicast-Border Gateway Protocol (M-BGP).
Key Solutions Offered.
Service Level Agreements (SLAs).
Policy--The Distributed Glue That Ties Everything Together.
Network Configuration Repositories.
Next Steps for DHCP.
Remote Authentication Dial-In User Service (RADIUS).
An Example of RADIUS in Use.
Introducing Distributed Policy Is a Large Challenge.
Lightweight Directory Access Protocol (LDAP).
Directory Enabled Networking (DEN).
Virtual Private Networks (VPNs).
Interactive Voice over Data Networks.
Voice over Data Challenges.
Deploying Interactive Voice over an Intranet.
Network Services Depend on Network Policy.
Real-Time Video over Data Networks.
Key Problems Addressed.
The Basics of Management.
The Sources of Network Management Data.
Engineering Problem Solving Applied to Network Management.
Switched Networks Challenge Network Management.
Network Management Needs of Switched Networks.
Proactive versus Reactive Management.
Delivery Guarantees and Traffic Prioritization.
Twelve Cardinal Rules of Creating Management Solutions.
Policy-Based Network Management.
A Conceptual View of Managing Tomorrow's Switching Solutions.
Management Strives to Be End-to-End.
Preparing for New Switching Solutions.
It wasn't all that long ago when Local Area Networks (LANs) were composed of Ethernet running on long segments of coaxial cable. In yesterday's LANs, several Personal Computers (PCs) and workstations shared the same cable and took turns using the network. As the segments grew beyond device and cable length constraints, bridges were added, providing an effective way to extend the number of devices and overall span of the LAN. Routers were used to access the Wide Area Network (WAN) and Internet. From the local router, data traveled across the Internet to its destination. The Internet was composed of a mesh of routers providing a communication infrastructure for moving data a few kilometers, many thousands of miles, or completely around the world.
Telnet, File Transfer Protocol (FTP), and other character cell interfaces provided both local area access and wide area access to the network. Graphical interfaces were used locally and often minimally.1 The network provided only limited communication, complementing the primary sources of information such as the telephone, face-to-face conversations, mailed reports, and nightly batch processing.
It is unclear whether the surge of networking began with the introduction of Microsoft Windows networking, the use of networked Graphical User Interfaces (GUIs), or just the need to distribute more and more data. Today's network growth could be due to the influx of the Web browsers and servers providing the ability to link multimedia (graphics, text, and sounds) with hypertext--after all, this was key to the World Wide Web (WWW) explosion. Or perhaps it was the penetration of client/server applications (sometimes classified as bandwidth "hogs"), distributed databases, e-mail environments, and file servers that was responsible for its proliferation. One other reason might be that technology has become much easier to deploy with the introduction of 10BaseT, twisted-pair wiring, and hubs. Most likely though, the popularity of networking came as a result of many, concurrent requirements and events.
Whatever the cause, the strain on the network gradually became very apparent. In the LAN, the standard 10 Mbps-shared Ethernet started to crumble, showing large periodic faults, consistent spikes of activity, and general sluggishness. More and more traffic on the Internet resulted in a bigger and bigger routing mesh. A larger mesh meant more routing change updates and more states of routing flux. In short both the LAN and the WAN were becoming unreliable and unbearably slow.
In the past few years we have seen radical changes in our networks: Bridges have been retired; there is less and less shared Ethernet in the LAN; the Internet is now composed of many Autonomous Systems (ASs) that are managed independently; and traffic is routed between the ASs at only a few external points in each AS. In short, our hunger for networking has resulted in a new generation of networking composed of technologies that scale to meet our needs.
Switching is the core to this new era of networking. Switches help networks scale by addressing performance and robustness and encapsulating network intelligence. In the LAN, switching provides the answer for bandwidth-hungry applications. Switching resolves the problems of shared networks by providing dedicated or minimally shared pipes between devices; and Ethernet switches provide dedicated 10 Mbps, 100 Mbps, and 1000 Mbps pipes, allowing for fat files, graphics, and remote data to flow without congestion. Switches bring together many layer-2 technologies including Fiber Distributed Data Interface (FDDI), Ethernet, Token Ring, and Asynchronous Transfer Mode (ATM), providing a great deal of flexibility for building LANs. It is no wonder that LAN switching became such an important technology so quickly and is so popular today.
Switching is also becoming the common denominator of the WAN. We are seeing more and more ATM switches that provide the needed capacity for moving large amounts of data quickly deployed in the WAN. ATM comes with the promise of Quality of Service (QoS) and a rich base on which to deliver that promise. Frame relay switches, commonly used to build corporate Intranets (private geographically dispersed networks), provide an effective way to connect LANs that are widely separated.
This book is about switched networks and the technologies incorporated within them--switching in the LAN and WAN, switching today, and switching tomorrow. Switches, including workgroup switches, backbone switches, access switches, edge switches, multiservice edge switches, and core switches, are the devices that are satisfying our insatiable appetite for more and more bandwidth. These switches span from the workgroup to the backbone in the LAN and from the edges to the core of the WAN; they are being combined with networking technologies to provide significant networking advances.
In short, switches are being positioned to take us into the next generation of networking. Layer-3 switching, layer-4 switching, multiprotocol label switching (MPLS), Virtual LANs (VLANs), 802.1Q, 802.1p, Class of Service (CoS), Quality of Service (QoS), Resource ReSerVation Protocol (RSVP), Internet Protocol (IP) multicast, Service Level Agreements (SLAs), and policy-based networking are all being actively introduced into switch architectures. These technologies are discussed in this book, as they are becoming part of our overall networking (switching) infrastructure.
Figure I.1 provides a conceptual view of the book. We work from the bottom up to establish our networking base and discuss current switching technologies in the LAN and WAN. Then (moving from left to right) we discuss technologies that are expected to shape tomorrow's switched networks greatly. We conclude with a discussion of networking policy and network management--the glue that holds our switched networks together.
This book consists of eight chapters.
In summary we start by defining our demands on networking and our network base in Chapters 1 and 2. We then cover switching today in Chapter 3. Chapters 4, 5, and 6 cover multilayer switching, guaranteed delivery, and multicast, respectively. Finally we look at network policy and advanced network management, which tie together tomorrow's switched networks.
Throughout the book it is assumed that you have some familiarity with networking, although time is taken up front to review some key networking concepts that are important to understanding the material in the book. The book assumes that you are pragmatic and interested in information that you can use to extend your existing set of networking knowledge and that you are after solutions and the motivations behind the technologies, not just a regurgitation of Requests for Comments (RFCs). For the most part, the book tends to stay away from the actual bits and bytes that go across the wire since this depth would require a book for each subject covered. Care has been taken to condense a lot of material into a few hundred pages that deal with the central theme of switched networks.
This book is written for corporate network analysts, network managers, information technology managers, network planners, network designers, technicians, and other technical management personnel who need exposure to these new technologies. It provides the necessary concrete information for planning an upgrade from a shared to a switched environment and a knowledge base of new, emerging technologies that are expected to be core to switching solutions carrying us into the next millennium. This book may also serve the engineering community or be used as a college text for a course in data communications.
If you look carefully at the list of people who will benefit from Building Switched Networks: Multilayer Switching, QoS, IP Multicast, Network Policy, and Service Level Agreements, you will extract one very common characteristic--all of these people are extremely busy! Like my last book, Managing Switched Local Area Networks: A Practical Guide, this book addresses this need by being concise and to the point, yet it is written in a friendly, easy-to-understand style. Books that can provide focused, pragmatic, integrated text are the most valuable ones for today's busy technical reader. This book attempts to be precisely that by focusing on some very interesting networking technologies, covering a lot of ground, and telling a complete story about switched networks.
As you read this book, you will quickly learn just how much information you need to know to understand the essence of networking. Over the past fifteen years I have been exposed to several network experts and many rich development environments. I have also been fortunate to go to many trade shows and external training events that provided a way to learn from experts in the industry. This book is a culmination of my experiences and knowledge, and it would not have been possible without the daily water-cooler talk, periodic brainstorming, critical thinking with colleagues, and exposure to the technical gurus across the industry.
My experiences at Wang Laboratories, Digital Equipment Corporation, 3Com, and, most recently, Nortel Networks are the basis for this effort. Although there are too many people to mention, I am very indebted to the technical community in which I work. I therefore offer a big thanks to all of my colleagues of yesterday and today.
Quality book writing has many critical components, and one vital component is first-class reviewers. I'm not completely knowledgeable of the process used to write books at other companies, but Addison Wesley Longman uses a series of very intense reviews during the writing process. These reviews make their books as technically accurate as possible, yet the books flow well. This book went through some very tough reviews by some strong network experts. I would personally like to thank Dr. Stuart Cheshire of Apple Computer, Dave Crocker, J. Alan Gatlin, Dave Hannum, Peter Haverlock, Glen Herrmannsfeldt, Mukesh Kacker, Jeffrey Mogul, Bob Natale of ACE*COMM Corporation, Radia Perlman, Linda Richman, Ed Volkstorf, and William Welch of Nortel Networks for their critical reviews and helpful suggestions and comments.
Last, but clearly not least, I would like to commend key members of the Addison Wesley Longman team who provided a tremendous amount of coordination, guidance, and support along the way. Writing a technical book is not a small task, and without a crew like this, you would see many sparse technical bookshelves. Many thanks to Karen Gettman, Mary Hart, and Maureen Willard for helping me pull this effort together into a book.
1. One notable exception is MIT's X Window System. This system is an early graphical interface noted for its voracious appetite of network bandwidth. Today the MIT campus uses a switched infrastructure.