Interconnections, Second Edition by Radia Perlman
Addison-Wesley, 532 pages plus index, $59.95
Review score: **** out of *****

If Yoda where to describe computer networks and the Internet he might say "The network is all around us. It connects us and through it flows information". Computer networks are becoming pervasive (like Yoda's "Force"). Computer networks are starting to become so familiar that most of us don't think much about them. They are just there delivering our data. Computer networks are also changing our idea of distance. As more high speed undersea fiber optic data lines become available, concepts of distance start to be erased as well.

Years ago I took a course in networking from Charlie Bass, one of the founders of an early networking company named Ungarman-Bass. At the time Ethernet was still considered pretty radical, the network bridge was new and Ethernet chip sets were just starting to appear. The networking course was held during the summer at UC Santa Cruz which remains the most beautiful University campus I've ever seen. After this enjoyable class I did not think much about networks except to welcome the increasing application of Ethernet. In fact, as networks became more and more a part of my life and work, I thought about them less and less. That is until I read Radia Perlman's excellent book Interconnections.

The full title is Interconnections, Second Edition: Bridges, Routers, Switches and Internetworking Protocols. Radia Perlman developed the bridge spanning algorithm which allows network bridges to connect networks without loops. Loops can result is packet explosion and network crash, so all bridges and some routers use a spanning tree algorithm when they join a network (e.g., at startup). Radia Perlman did her Phd work at MIT on networking (Network Layer Protocols with Byzantine Robustness, MIT 1988) and has been involved in network and protocol issues for many years.

Computer and data networks are increasingly pervasive and are a huge and rapidly growing industry. So far there is an insatiable demand for network bandwidth, which means faster networks and network routers. The importance of computer networks and their technological demands make them inherently interesting. But the actual operation of networks can be obscure and full of strange acronyms like TCP/IP, UDP and NAT. Having written socket level client server software I knew about some of the higher level networking issues, but I had only a vague understanding of how routers and network bridges worked.

In the hands of an academic writer, a description of routing and network protocols could be bone dry and boring. Fortunately, Radia Perlman is not only a recognized expert in networking, but an excellent writer. Years of working on network design and taking part in design committee meetings seems to have given her a rye sense of humor. Or perhaps she has survived all those standards meetings because she had a sense of humor and irony to begin with.

One of the questions I had before I read Interconnections was: "Why does Cisco have such a huge market share?" I am not a networking expert yet, but after reading Interconnections it appears that the answer is that although the networking and routing standards define basic protocols and operation, there are many undefined areas, particularly on network startup and configuration (e.g., when a router joins a network, for example). Cisco has a particular implementation that fills these holes (otherwise the network would not work). However, they do not disclose how they implemented the features that are not clearly defined in the standard. Other router vendors must reverse engineer Cisco operation in an attempt to produce compatible hardware. On occasion other routers do not work properly when connected to a Cisco network. This encourages people to buy Cisco hardware.

Although computer networks based on TCP/IP are twenty years old, networking and routing remains remarkably primitive. Although networking remains in an early stage of development, the field is growing at a rapid pace. Radia Perlman's Interconnections provides an excellent starting point. The only flaw I found in the book is that it lacked a bibliography or "suggested reading" list. Some references are given as foot notes, but there are hard to find after putting the book down.

Ian Kaplan - 5/2000

Book review table of contents

back to home page