A New TCP/IP Classic 97
FrazzledDad writes "Network geeks and developers working in the TCP/IP domain are most likely familiar with Douglas Comer's Internetworking With TCP/IP Vol.1. Comer's book was central for my understanding of how things really worked in the small corner of a world-wide network I use to manage. Charles Kozierok's The TCP/IP Guide has knocked Comer's book off my shelf. Kozierok's weighty book (1600 pages!) does a terrific job both as a reference and as a learning aid." Read on for Jim's review.
The TCP/IP Guide: A Comprehensive, Illustrated Internet Protocols Reference | |
author | Charles Kozierok |
pages | 1616 |
publisher | No Starch Press |
rating | 9/10 |
reviewer | Jim Holmes |
ISBN | 159327047X |
summary | Amazing broad, deep coverage of TCP/IP in an understandable fashion. |
Kozierok spent at least four years working full-time on this book, according to the dedication, and it shows. He covers everything from networking fundamentals to individual application protocols such as Gopher.
Do you need to familiarize yourself with Open Shortest Path First (OSPF) routing protocol basics? It's covered. Do you need to understand the pros and cons of Network Address Translation, and how static and dynamic mappings work? It's covered. Do you want the nitty gritty of how message formats are laid out? It's covered.
Kozierok also presents several chapters specifically on IPv6, laying out changes in the new version before diving into the nuts and bolts of it. He discusses the major additions, and dedicates an entire chapter to the new addressing scheme. The transition from IPv4 to IPv6 is a well-written section talking about the difficult conversion between the two versions.
THE BOOK AS A LEARNING GUIDE
TCP/IP can be a rather dry topic to read about when trying to learn portions of it. Let's face it: reading about BOOTP's messaging over UDP is not something most folks will give up a Friday night on the town for. OK, Kozierok's writing style won't make that happen, but he does keep things interesting and flowing well enough that working one's way through such topics is actually entertaining instead of torture.
For example, Chapter 18's discussion of subnetting concepts lays out the fundamentals in clear order without sliding into unfathomable academic blabberspeak. His use of "Key Concept" boxes throughout the book helps point out important items.
Just as important to the book's clarity and usefulness are the amazing graphics. In the Acknowledgments Kozierok specifically thanks the folks at SmartDraw.com for their illustrating package. He's put the tool to fantastic use for everything from breaking out the control bits from a TCP segment header to showing how iterative DNS name resolution works.
THE BOOK AS A REFERENCE
The level of detail in the book makes it a valuable reference in addition to its role as a learning guide. For example, readers can find specifics on details of SNMP data types, NFS server procedures, or TCP segment format layout. Additionally, Kozierok discusses many of the various TCP/IP utilities, such as using "netstat" for troubleshooting with a detailed discussion of various outputs.
Kozierok must have spent a lot of time figuring out how to best lay out the book, and it pays off with sensible organization. Two tables of content, one brief and one detailed (32 pages!), help one to get to the right spot to look up needed information. The index is nearly 50 pages and seems to be quite exhaustive; another great tool for getting to the right spot. There are also comprehensive lists of Figures and Tables if you're trying to access something via that route.
WHAT IT DOESN'T COVER
Kozierok is upfront about things he's left out of the book. You'll need to look elsewhere (back to Comer's book, perhaps) for details on TCP/IP in ATM networks, security and firewall design, and the lower levels of socket usage.
CONCLUSION
To me, a significant advantage of this book is No Starch's binding system that they make so much hay about. I can open this massive book to any point and leave it flat on the table. Pretty impressive!
Kozierok also has a companion website (www.TCPIPGuide.com) with errata, a FAQ, and various other areas. You can also purchase an electronic copy of the book.
The TCP/IP Guide is a tremendous work, and it's a significant resource for anyone working with TCP/IP."
You can purchase The TCP/IP Guide: A Comprehensive, Illustrated Internet Protocols Reference from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
Been done before (Score:5, Insightful)
Theres only so many books you can read on the subject - no matter how hard core you are.
And the number of those books is... (Score:2)
Re:Been done before (Score:3, Insightful)
Re:Been done before (Score:4, Insightful)
This book is pretty good. I'll be using it to teach a class in the spring. Its not perfect but the best I've found.
Re:Been done before (Score:1)
you are wrong (Score:5, Insightful)
No, dude. We read "TCP/IP Illustrated, Volume 1" by Richard Stevens. The bible.
Re:you are wrong (Score:2)
I think I see a /. poll coming...
Save THIRTY ($30) BUCKS! (Score:3, Informative)
Re: (Score:2)
Bookpool has it for $49.95 (Score:2)
Re:Save THIRTY ($30) BUCKS! (Score:1)
Re: (Score:2)
Re:Save THIRTY ($30) BUCKS! (Score:1)
It's not that the most of the moderators are shit, either - I'm sure that's not the case.
I believe it's because the whole M1 system is fundamentally fucked up. The most boring of points go from Score 1 to Score 5 immediately due to people looking at pages that are 30 seconds old, and moderating them. Then the "Underrateds" hit and drive things b
But is Comer actually that good? (Score:4, Insightful)
However, in a phone interview recently, I was told that the tear-down on a TCP/IP session was a four-way handshake. Websites I was pointed also said this. But when I go to the Comer, Vol. 1, it says that it's actually a six-way: a three-way from the originator, and a three-way from the recipient.
Which, of course, leads me to wonder about his accuracy.
mark "and the O'Reilly TCP/IP book says three-way...."
Re:But is Comer actually that good? (Score:5, Informative)
IPv6 ??? (Score:5, Funny)
We just finished upgrading to IPv5, and it cost us nearly $1.2 million
That's the last time we hire "Dogbert IT Services"
RIP Jon Postel (Score:2, Interesting)
Why wade through someone else's interpretation when the specifications themselves are so good and freely available?
Re:RIP Jon Postel (Score:5, Funny)
Re:RIP Jon Postel (Score:2)
big bellies, living in their mothers basements ?
Re:RIP Jon Postel (Score:2)
... and RIP W. Richard Stevens. (Score:2)
Save $20 (Score:5, Informative)
Re:Save $20 (Score:2)
Buy direct from the guy (Score:5, Interesting)
Why not kick back a few bucks his way to reward him for his good work?
Or Free Online! (Score:2, Informative)
1600 pages? (Score:5, Funny)
Re:1600 pages? (Score:5, Funny)
My theory was that it wasn't bolted to the table to keep people from taking it, but rather to keep it from falling off and killing someone when it landed on them. I can just picture the legs of some poor freshman sticking out from under that paper monster, the rest of him having been squished into jelly from the book's sheer weight...
Re:1600 pages? (Score:4, Funny)
Re:1600 pages? (Score:2)
If it were closer to the business majors, I might have had the theory that the book was bolted to the table to make it more difficult to beat MBA's with =]
several linear feet? big deal (Score:3, Informative)
Now *those* were manuals.
Danged useful, too.
Still the standard, in my book.
As much as I love *nix, it still falls way short
of VMS in a couple of areas, and one is good documentation.
Re:several linear feet? big deal (Score:2)
Oh, and saying *nix falls short of VMS in a couple of areas is like saying DOS falls short of *nix in a couple of areas...
Re:several linear feet? big deal (Score:1)
Re:several linear feet? big deal (Score:2)
Re:1600 pages? (Score:1)
RoR (Score:3, Funny)
From what I have read around, the Ruby manual is so hefty that lots of people have Ruby on rails.
1600 pages? (Score:2)
/The last time I read over a 1,000 pages from one author, it wasn't worth it.
Re:1600 pages? (Score:1, Funny)
Re:1600 pages? (Score:3, Funny)
SiO2
Re:1600 pages? (Score:1)
That's a shame. I quite enjoyed the Baroque Cycle.
Re:In case you missed the first 10 comments (Score:1)
better- Stevens...Unix Network Programming Vol.1 ? (Score:2)
My main complaint with those books the example code uses his own libraries. Which is fine for brevity but can be a little confusing. Also not the easiest to learn from, but great when you know enough.
Re:In case you missed the first 10 comments (Score:2)
Gopher?!?! Now there's ancient history... (Score:3, Funny)
And is coverage of Gopher supposed to be a selling point for this book??
I haven't used Gopher since... well, actually, I never used Gopher for anything other than idle curiousity to see what the hell it was. The Web made Gopher completely obsolescent.
Talk about a way to pigeonhole your book as "old news".
Coverage of bittorrent would be far more interesting and relevant.
Re:Gopher?!?! Now there's ancient history... (Score:2)
Re:Gopher?!?! Now there's ancient history... (Score:2)
Re:tcp/ip is a Good Thing (Score:2)
the only one commonly available off the shelf is the not-so-humble Sharp Zaurus.
www.oesf.org for more information
Re:tcp/ip is a Good Thing (Score:2)
Re:tcp/ip is a Good Thing (Score:2, Insightful)
http://www.conklinsystems.com/palm/online.php [conklinsystems.com]
as in beer (Score:3, Informative)
Re:as in beer (Score:2, Interesting)
Not only that, he answers his own email. (Score:5, Interesting)
I've read a LOT of networking crap over the 25 years I've been doing computer networking, this ease of reading is not common to the genre.
His work is also online:
http://www.tcpipguide.com/free/index.htm [tcpipguide.com]
Bob-
Why do I need a book? (Score:3, Interesting)
I guess what I'm thinking is that TCP/IP networking is hardly rocket science. Surely the basics can be described in just a few pages. For everything else, you're going to have to look at specs anyway.
Am I missing something? Or are network prototcol programmers jealous of the multi-thousand-page-red-softcover-with-the-author
Re:Why do I need a book? (Score:2)
1. 1600 page books look impressive.
2. Keep the information hidden. bury it in useless data and it is hidden in plane site.
3. Make TCP/IP look 1000x more complicated than it is. Network Engineers get paid more.
1600 pages for something that would have to have a lot of fluff to fill half that and could probably be covered really well in 1-200 pages.
ok smart guy, let's see it (Score:1)
Re:ok smart guy, let's see it (Score:2)
SNMP data is organized in a tree-like hierarchy with numbers assigned by company, type, and sub-types as necessary.
Examples:
Well, that's a good chunk of what you need to know. Examples would deal with the rest practically. PS, like a previous poster said, the excessively detailed stuff belongs to the RFCs.
Re:ok smart guy, let's see it (Score:1)
Re:Why do I need a book? (Score:4, Insightful)
For my thesis I described the basics of TCP/IP in about 20 pages; however, when onse is making changes to the TCP stack itself, the basics just will NOT do. I have hit several of the in-dept chapters of the Steven's volumes multiple times. Some times I find those books lacking in the detail I need.
These are not "admin" style books. Most TCP/IP protocol suite implementations are very large and complex.
28.8kbps modem? (Score:2)
About 2.5 hours using a 28.8kbps dialup modem; about 1.3 hours at 50kbps; about 8 minutes with 500kbps broadband." (http://www.tcpipguide.com/faq.htm [tcpipguide.com])
Has anyone seen a 28.8kbps modem lately? I thought they were extinct.
Re:28.8kbps modem? (Score:2)
I also "see" them a lot on older computers here, technically...But I'm guessing you meant see them being used.
Re:28.8kbps modem? (Score:2)
Re:28.8kbps modem? (Score:2)
Re:28.8kbps modem? (Score:1)
Oh ya, if you live rural, it's a real bother to try and find "broadband" in the US. It just ain't happening except for very expensive and very limited satellite "service". Cable is unobtanium and DSL means you have to be two miles or less from a telco box, which barely qualifies as suburban, let alone "rural". I think when/if Wi
Re:28.8kbps modem? (Score:2)
TCP/IP Illustrated has been a de facto reference (Score:2, Informative)
For many years until a RFC was published, the "official" reference (for example, to quote in an article or book) about TCP's fast retransmit and fast recovery was Stevens' book, unless you wanted to quote the original Usenet post for Van Jacobson.
Did this book cover QoS (Score:1)
Free IBM redbook is great (Score:5, Informative)
thank you IBM - its a fantastic resource and reference.
Re:Hypertext learning. (Score:2)
Yup - we do. And only time will tell us whether ink on pulped wood is a better medium for reference than electronic media, or whether us old-timers just like it by habit
an inte
knocked Comer's book off my shelf - why? (Score:2)
Oxymoron? (Score:1)
From the Oxford American Dictionary:
classic |klasik| adjective judged over a period of time to be of the highest quality and outstanding of its kind.
Heck, it has been almost 3 months.
Comment removed (Score:3, Funny)
No, network professionals are (or should be) (Score:3, Informative)
Re:No, network professionals are (or should be) (Score:3, Informative)
The second edition of Volume 1 has just been released.
Volume 1, and do more than cover just TCP/IP standards, they also have a practical implementation aspect. Of course, its all Cisco based, but given a huge whack of the networking is as well, that's not as big a disadvantage as you'd expect. For BGP (especially Cisco's implementation of BGP), look to the CiscoPress "Internet Routing Architectures". I prefer it to the coverage in Volume II of Routing TCP/IP.
However, you will be g
Read it (Score:1)
The gold standard is still "TCP/IP Illustrated", Stevens, even though it is getting somewhat long in the tooth. The Kozierok book is essentially all prose descriptions of how the TCP/IP stack works, no code. There is simply no replacement for Stevens if you need to figure out how the stack is dealing with multicast packets or dozens of other situations.
There are some questionable organizational choices, such as starting of with SLIP and PPP in the first chapters. And gopher is pr