 |
|
 |
|
Comer's book has 32 chapters broken up into 4 sections:
Tanenbaum's book is broken up roughly the same way, with each chapter being very large and equating to Comer's sections:
Comer in the preface tells who the book is for: a one or two semester undergraduate course where the goal is breadth, not depth. But most student's objective is to learn how the internet works: this entails an understanding of how networks work, but on a somewhat superficial level.
Comer re-emphasizes his points with a short italicized paragraphs prefixed with to summarize spread throughout the text. I suppose these to summarize blocks come from Comer's personal experience of which points to emphasize to students. I would find them useful if I was studying the material for a test.
I found Comer's discussion of the programs ping and traceroute informative and useful. I also liked the discussion of fragmentation and packet reassembly. But as an experienced network programmer, they were superficial (concentrating on the concept, rather than the implementation).
Comer's book is not intended for electrical engineers, but people who want to learn how networks work. In Tanenbaum, physical mediums are discussed in depth, with a good discussion of topics such as Manchester encoding, wireless communications, the telephone company and how modems work, ISDN and ATM and 10 Mbit versus 100 Mbit ethernet. One of the things Tanenbaum does extremely well is put all these things into a historical perspective, giving background into how these things evolved (in addition to the technical aspects of how they work, along with numerous references).
Tanenbaum is full of wry (and sometimes side-splitting) humor. For example, there is a saying in the book when he discuss network carrier bandwidth:
"Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway." -- page 83
Such advice is both humorous and practical.
Comer has a chapter (12 pages) on the Berkeley socket API. Tanenbaum mentions the socket API briefly (1 page). From experience, if the reader wants to write source code, a more complete description is needed with examples (these are available in Comer/David Stevens' Internetworking with TCP/IP Volume III, Prentice-Hall, 1993 or Richard Stevens' Unix Network Programming, 2nd Edition, Prentice-Hall, 1997).
Both books cover a full range of network applications (DNS, email and SMTP, SMNP, http and java). Comer also covers basic protocols like ftp. They both cover network security, but Comer deals with it in about 5 pages, while Tanenbaum gives it almost 50 pages (giving a good elementary lesson in cryptography). I'm personally weak on cryptography and network authentication protocols (I still need to read Schneier's Applied Cryptography) Comer didn't touch on anything I didn't know about cryptography, giving a cursory overview of the topic. Tanenbaum presented the information in detail, discussing attacks and solutions in an informative and entertaining way.
Comer provides a good glossary and a 9 page bibliography. Tanenbaum provides a 20 page bibliography, and a good discussion of literature available in the field. A glossary would have been useful in Tanenbaum's book, there were a few acronyms I didn't previously know which weren't in the index.
In addition, Comer provides a CD-ROM, but I wonder how useful it is. Most of the CD are the images shown in the figures of the book. In addition, there are packet traces of binary data files from the Solaris tool snoop (which appears to leave it as an exercise to the reader to write a tool to decode them). With all the space available on the CD-ROM, I'm surprised ascii files aren't distributed. In addition, I feel it would be more useful to provide tcpdump traces (which has been around for 7 years (from ftp://ftp.ee.lbl.gov, runs on a large variety of Unix systems, and was even used for examples by Richard Stevens in TCP/IP Illustrated Volume 1. With all the space available on the CD-ROM, it could have been used to distribute the entire RFC collection (which aren't mentioned in Comer's text) along with sample programs. The contents of the CD-ROM also appear to be on the web at http://www.netbook.cs.purdue.edu.
Comer is a fine book for a course, but it will not suffice as a reference book. It is focused as a textbook on a specific course. As a consumer, I thoroughly enjoyed his Prentice-Hall title The Internet Book, which is smaller, much cheaper and is not a textbook (it is intended for curious internet users). Tanenbaum's is an excellent reference book (and might be somewhat intimidating for course material, there was significant increase in size between the 2nd and 3rd edition). I wouldn't hesitate to recommend Tanenbaum for one's bookshelf or engineering course work. Comer presents a smorgasbord approach to networks, Tanenbaum is a catered, seven course meal.
-- Marty Leisner (leisner@sdsp.mc.xerox.com)
| Readability |
|
| Originality |
|
| Organization |
|
| Accuracy |
|
| Consistency |
|
| Depth |
|
| Timeliness |
|
| Editing |
|
| Design |
|
| Overall Value |
|
| Readability |
|
| Originality |
|
| Organization |
|
| Accuracy |
|
| Consistency |
|
| Depth |
|
| Timeliness |
|
| Editing |
|
| Design |
|
| Overall Value |
|
Explanation of ERCB rating scale: No stars = unacceptable, 1 Star = marginal, 2 Stars = average, 3 Stars = above average, 4 Stars = exceptional.