Browse Books

Reviewer: Jeff D. Parker

The trade press is full of articles on the emergence of the open systems interconnection (OSI) suite of communication protocols and the demise of the existing TCP/IP protocols. These articles always talk about the benefits of OSI and the ubiquity of TCP/IP. They go on to discuss the difficulties vendors have encountered in delivering promised products based on OSI (OSI has been two years away from deployment for some time) and the importance of positioning in the coming shakeout. Most of these articles are long on marketing insight but short on technical detail. While the marketing perspective is important, a clear exposition of the technical issues is needed. Although the standards are available, they are long, expensive, difficult to understand, and often incomplete or even obsolete. Rose's book is a long-needed guide to this important area. It is a detailed discussion of OSI from an engineer who has been intimately involved with the suite. It is unusual to find an engineer who is so involved in the process and has the time and the energy to produce a manuscript of this size, but Marshall Rose is an unusual engineer. Those who have read /mtr's notes to various Internet news groups will recognize the author's voice: at once precise, authoritative, and a bit peeved with a world that does not meet his standards. Rose does not enjoy the endless politics of standards negotiations: he leaps in to implement. He has produced, with the aid of a few like-minded souls, several Internet proposals (RFCs) and the ISO Development Environment (ISODE), a partial implementation of the upper-layer OSI protocols that runs over TCP/IP. The ISODE source is available to all for a nominal charge, and it has been widely distributed and used. While few companies are willing to admit that their software is based on a free product, it is suspected that more than one offering is based on the ISODE.<__?__Pub Fmt eos-space>The text covers the technical issues behind the OSI; the ISODE, as a proof of concept; and the politics of standards. The discussion of OSI includes a systematic voyage through the seven-layer ISO model, starting with the network layer, level 3, and going up through the application layer. Readers who wish to learn more about the first two layers (physical and data link) are referred to Tanenbaum's now-classic text [1]. Rose presents the major primitives and protocols of each layer and places them in perspective. He does not hesitate to point out problems with the standard, using an unusual textual convention. “Soapbox” icons in the margin mark passages that are highly charged. These symbols allow Rose to vent his opinions freely, while placing the reader on notice that the material is controversial. Although the rest of the text is not free from Rose's passions, these markers provide a useful way of allowing him to vent his spleen, while preserving the reader's forbearance for the rest of the book. Rose uses the soapbox to denounce technical problems with OSI (such as the mismatch between the OSI transport services) and as a pulpit to excoriate others in the business. A singular contribution of the technical portion of the book is an extended discussion of the Abstract Syntax Notation (ASN.1) and the Basic Encoding Rules (BER). These allow an OSI application to describe an arbitrary data structure and transmit it over the wire. While similar protocols are in use, such as NDR and XDR, none have the power that ASN.1 provides, and none require so much elucidation. This section ends with a discussion of three OSI applications that provide general versions of name service (X.500), mail service (X.400), and file transfer (FTAM). Along with the network management application (CMIP), these promise to be the initial OSI offerings. The second aspect of the book is a discussion of the ISODE. The discussion is not detailed enough to help the user of ISODE, who is referred instead to Rose's extensive User's Manual, nor are these sections broad enough to help those who do not have access to ISODE. They may encourage more people to obtain the ISODE and experiment first hand. The final aspect of the book is a discussion of the organizations and politics of the communications standards community. While standards exist for objects like light bulbs and color television, the number of different bodies and the scope of activity in computer science are unprecedented. In part, this activity is due to the nature of computer software, which admits endless proliferation of detail, and in part it reflects the long history of proprietary solutions to universal problems and the difficulty of simultaneously weaning vendors, suppliers, and clients from these partial solutions. This leads to a process of negotiation in standards bodies and to perpetuation of artifacts of long-departed equipment in modern systems, like the coal stoker in the diesel cab. Bismarck's comment about sausage and legislation applies to standards: if you wish to enjoy them, you should not observe them being made. Rose has little patience for this process. Here is a small sample of his ire: “Vendors send persons they can afford to have out of the office, the professional standards Goers. The Goers have likely never implemented anything in their professional careers.<__?__Pub Caret>…A Goer attends meetings to deal with weighty political issues, travel, and have many fine lunches and dinners.” The book's appearance has problems: it is a victim of the proliferation of desktop publishing systems. Excessive white space, grammatical errors, and punctuation errors mark this as a book that did not receive enough editing. Had it gone through the traditional publishing channels, however, it would not have emerged so soon. We can hope that Prentice-Hall does a better job with the second edition. Others will find the author does not have enough reverence for august bodies and distinguished engineers. This, of course, is much of the book's charm. Anyone wishing to learn about OSI, and anyone involved in communications protocols, should read this book. It is a spruce in a stand of dwarf pines. Oh, and those lunches and dinners__?__ They're not really that fine.