Browse Books

Reviewer: Arthur E. Salwin

The Internet is a dynamic place. Links continually go up and down, and new nodes are added daily. The routers, which forward the packets from one host to the next, must track these changes and update their internal tables to reflect the current network configuration. To exchange the necessary information, router-to-router protocols are used. These routing protocols are the subject of this book. The book begins by discussing IP—the Internet Protocol—upon which all the other protocols are built (“IP Over Everything”). Next, the book describes two families of interior gateway protocols. In the distance vector family, information on the number of hops to other routers (that is, “distance vectors”) is exchanged among the routers; in the link state family, each router maintains a map of the entire network. RIP, a distance vector protocol, was originally used on the Internet. It has the advantage of relative simplicity, but problems inherent in it have caused it to be supplanted by OSPF, a link state protocol. As the Internet grew, it became apparent that maintaining a single monolithic network was not feasible, so it was divided into a set of autonomous systems (ASs). The interior gateway protocols maintain the routing tables within an AS, but a new class of protocol, the exterior gateway protocol (EGP), was needed for the ASs to exchange routing information. EGP's limitations have led to the Border Gateway Protocol, which must be used in conjunction with the Transport Control Protocol (TCP) of TCP/IP fame. The last section of the book discusses several functions that are not currently well supported by the Internet. These are fueling the development of new protocols. The first such function is multicast, which has obvious applications for disseminating information to multiple users. But multicast also has applicability behind the scenes in keeping the routing tables up to date. Another needed function is the support of mobile users—whether they are traveling in a moving vehicle or merely accessing the Internet from a stationary laptop computer at a new location. To accommodate them, Internet addresses must be divorced from the topology of the network; several proposed schemes for doing this are discussed. Finally, the book addresses a potentially serious problem—running out of Internet addresses—a situation whose prospect would have seemed ludicrous and probably unimaginable to the Internet's original developers. The Internet's rapid growth means that 32 bits soon will not be enough to specify a unique address. The book is optimistic that some interim steps will suffice until longer- term, more permanent solutions are implemented. Huitema presents a technically difficult subject in a clear fashion. Concepts are clarified by repeated use of a simple six-node network. Step-by-step updates of the routing tables for this example network are shown as various algorithms are applied to it. The chance of typographical errors in presenting such material is high, yet the book is remarkably free of them. This book is a serious text and is not written in the chatty, informal style that is common to many Internet books. That is not say it is a dry reference about bits, bytes, and headers, however. Mild humor is employed throughout, and Huitema makes the information relevant by presenting routing algorithms in the context of their historical setting and their performance implications. The book even discusses some concepts that did not stand the test of time, and tells why they did not. Among these are centralized routing and having the routing tables respond to dynamic loads. If you already know how to access the Internet and now want to learn more about how the bits get from here to there, you can learn a lot by reading this book. I know I did.