Open Systems Networking: TCP/IP and OSI D. M. Piscitello & A.L. Chapin ISBN 0-201-56334-7 1993, Addison-Wesley Publishing Co. Reviewed by: D. Crocker, 5 Jan 93 With the recent explosion of interest in the Internet has come an extraordinary profusion of books about its technology. One of the most recent offerings is by two luminaries, whose OSI credentials date back to its start. Messrs. Piscitello and Chapin provided seminal contributions to the OSI effort, most notably in its architecture and lower layer protocols. More recently, they have become my colleagues on the Internet Engineering Steering Group which oversees Internet standards development. Hence, the authors have a depth and breadth of experience that should bode well for analysis and perspective. By and large, that is the major strength of this book. The book is extremely ambitious. It provides descriptions of architecture, protocol details, comparisons between OSI and TCP, and description and commentary about standards making, all in an extremely readable style. Small sections are marked "AHA" to warn the reader of personal commentary, but the main text goes beyond simple exposition, too. The first seventy pages are devoted to discussion of architecture and concepts, most of which were quite well done though I have never thought of ASN.1 as a "language". In any event, anyone who devotes a distinct chapter to discussing naming and addressing gets extra points from me. As expected, the content covering the lower layers of OSI is excellent. The book is rich in detail about the specification effort and has an extensive chapter on routing. They also provide constant commentary about the choices and difficulties in creating the standards. Often this includes very pointed criticisms for the choices made; such candor is rare and welcome. Discussion of OSI's upper layers is notably weaker in assorted technical details, but is well organized, thorough, and makes for a good introduction. Some people assess a French restaurant based on its rendition of onion soup which is a relatively simple dish, ruined remarkably often. I like to do a quick check of a TCP/IP book by reading their description of the TCP Urgent data mechanism which is highly idiosyncratic. The authors do better than average and get credit for providing quite a bit of detail; unfortunately key portions are wrong. One problem also occurs in a few other places in the book: a slight tendency to mix discussion of implementation with discussion of protocol without distinguishing them. Overall, the sundry difficulties with details of the TCP/IP stack and its standards process mean that this book should not be used as a first-encounter with that technology, but it is fine for follow- on reading.. Given that TCP and OSI remain major players in the life of data networking professionals, there is great benefit in being able to read about them in a combined text, particularly when the discussion includes the authors' opinions about the two stacks. It is curious that so much of the commentary freely and frequently admits to the various failings of the OSI technology, but repeatedly and inevitably concludes that its future remains bright. I'm tempted to suggest that the authors are looking at things through Rose colored glasses, but Marshall might not agree. In truth, this is an OSI book with a great deal of TCP/IP content. Some books should be read as part of a one-way communication process, in which the author conducts a monologue, filling the reader's brain with new facts and figures. Other books work best as a two-way discussion between the author and the reader, the author triggering reactions of surprise, agreement, disagreement, and further thought. Open Systems Networking is such a book. If you count yourself as a professional in data networking, you need to develop a broad perspective on both architecture and protocol details. This book will help you.