BKENCNTW.RVW 20000114 "The Network Press Encyclopedia of Networking", Werner Feibel, 2000, 0-7821-2255-8, U$84.99/C$127.95/UK#60.99 %A Werner Feibel %C 1151 Marina Village Parkway, Alameda, CA 94501 %D 2000 %G 0-7821-2255-8 %I Sybex Computer Books %O U$84.99/C$127.95/UK#60.99 800-227-2346 Fax: 510-523-2373 %P 1444 p. + CD-ROM %T "The Network Press Encyclopedia of Networking, Third Edition" Writing an encyclopedia is a difficult job, no question. It must be particularly difficult in a technical field. Feibel has obviously put a lot of work into the project, but the result remains problematic. First off, it is rather difficult to see this as an encyclopedia. There are a great many short entries simply defining terms, so the book might be closer to a dictionary. There are, though, a number of longer articles on major topics. The second point to make is that not all of the book is about networking. Granted, it is difficult to say where to draw the line between technologies, but a great number of listings refer to computers, particularly of the Wintel/PC variety, and have little or nothing to do with networking or communications. On the other hand, "AI" refers only to authentication information, with no mention of the rather more well known artificial intelligence. The original title was "Novell's Encyclopedia of Networking," and that still shows up in entries such as "Access Rights," where the material is completely NetWare specific. "//" is defined (Novell owned the UNIX trademark for a while) but not the Microsoft equivalent "\\." However, there is a rather good piece on the Windows NT Administrator account, among others, so Microsoft is by no means ignored. Some articles have a depth that is hard to find even in specialized books on the topic. For example, I have reviewed texts dedicated to firewalls that only describe packet filters, with no mention of proxy servers, let alone the two different types. There is an excellent essay on application proxy servers (albeit with lousy examples) in here, but it is followed by two rather shoddy pieces on circuit level proxies and firewalls respectively. And that, unfortunately, seems to be a rather big problem. For every good bit, there are several parts that are misleading, poorly explained, or flat out wrong. Some mistakes can be put down to pure carelessness, such as calling Corel "Lerel," or Teledesic "Teledisc." Other times the wording or explanation is negligent, such as the assertion that, in 7-bit ASCII, the eighth bit is used for parity. (This depends entirely on the situation.) Bang path addressing seems to be conceptually understood, but poorly illustrated, whereas it is hard to say whether the concept of "store and forward" is understood at all. It is difficult to see how listings like "DS" (as in the bandwidth levels of DS-1, DS-3, and so forth) explain anything. And I'd defy anyone to justify the definition of HTML (HyperText Markup Language) as a scripting language. The article on 56K modems has a number of errors, and even a logical fallacy. The discussion of agents makes no distinction between viruses and mobile code. (On the other hand, Fred Cohen might like that.) "Algorithm" contains a rather odd grab bag of examples. Lots of words and examples still fail to properly explain either the complete function or the usage syntax for anchor tags. The description of an antivirus confuses the various types of antiviral software with modes of operation. The entry for archie isn't too realistic, and is probably dated. The illustration for graded index fibre optic cable is completely backwards. "Hit" makes no reference to Web sites. Part of the problem is that Feibel seems quite willing to include his own, or at least very non-standard, terminology. "Cathedral" is used to refer to proprietary software, and, while Eric Raymond's piece on "The Cathedral and the Bazaar" is very good, I'm sure that even Raymond would agree that "open source" is more widely understood than "bazaar." "Optimistic security" is fairly easily construed, but it is not a term that is used in the security field. Since the entry for "typewriter" is obviously a joke, you have to peruse the Jargon File to find out that somebody wasn't having Feibel on about "bytesexual." Many extremely specialized terms get very brief entries that don't explain much. Ordering of the numbers section goes by the size of the number, not alphabetic ordering, so that 802.2 comes before 1000 which comes before 3780 which comes before 6611 which comes before 41449. Cable refers only to twisted pair, except that there are also separate listings for "cable, coaxial" and "cable, fiber-optic." Many protocols are not listed as themselves but as "protocol, ...," and, combined with the format for cross references, this appears to make "CHAP" refer to "Challenge Handshake Authentication Protocol" which refers to "CHAP" without ever telling you what it is. Cross references are also spotty: ADSL (Asymmetric Digital Subscriber Line), DSL (Digital Subscriber Line), SDSL (Symmetric Digital Subscriber Line), and VDSL (Very-high-speed Digital Subscriber Line) don't refer to each other, and none refer to HDSL (High-speed Digital Subscriber Line)--which refers to them all. The article on ActiveX is good, reasonably fair and complete. The definition of freeware is much better than in most dictionaries. Instant messaging is right up to the minute (as opposed to PGP, which hasn't been updated since the second edition, and BITNET which was probably out of date when the first edition came out). The listing for viruses is much better than I have come to expect. Overall, however, the work is simply not as reliable as one needs an encyclopedia to be. It might be handy as a reference to trigger a reminder, but if you don't already know the technology you cannot be sure that what you find here is the straight goods. copyright Robert M. Slade, 2000 BKENCNTW.RVW 20000114