Customer Reviews

TCP/IP & Linux Protocol Implementation: Systems Code for the Linux Internet

5 star:		(4)
4 star:		(0)
3 star:		(1)
2 star:		(3)
1 star:		(6)

 

 

My opinion is this book is pretty much just some relatively high level networking concepts and a printout of a bunch of the Linux kernel source code.

I guess some value add is supposed to be in giving some idea of which code fragments in the kernel source files do what, but I do not think it does a very good job of that, since it does not start with a top-down description of what needs to be done and then proceed towards more detailed views, finally ending up with annotated source code. It pretty much just gives the top level, and page after page of printouts of Linux kernel source.

For me, this adds very little value, so the book is worth very little to me. At the very least I think anyone considering buying the book should take a look at it before spending money.

I was cheated into believing this book might have a lot of meat in it because of number of pages (almost 1000 pages), but I was wrong when I started reading the book. Over 90% of the pages are kernel source code, literally, without any word addition or deduction. It might just include a CD ROM which contains the source code and make this book a better buy at 100 pages (if charge less also).

This book wants to teach you tcp/ip protocol and try to sell you on the Linux front since it is hot today. But it fails in both aspects. The tcp/ip protocol descrition is so short and incomplete that I don't believe anyone reading this book along would become familiar with tcp/ip. Nor does this book teach you about how to understand/modify kernel tcp stack in case you want to do something your own.

It is a shame that Vinton Cerf is the editor of this book and gives comment comparing that to "TCP/IP Illustrated". This book is way below the league and looks like some college students finishing a school report (no time, let's fill up with copied materials!). Vinton may lose his sense of technicality due to MCI Worldcom bankruptcy.

Finally, I would comment on the organization of the book. Each chapter starts with some summarization of introduction, then copy of relevant kernel code, usually in sections with one line description from the author. I give the book two start mainly for these brief introduction and summarization.

save your money.

90% of the book is kernel Source Code without almost no explanation. Why buy this book if you can view the source code for free

this book is just an imprint of the kernel
with under 0.1% of documentation. do not buy it.

in linux-netzwerkarchitektur from wehrle, paehlke, et.al.
and device drivers from rubini, corbet you will find better
better and more detailed descriptions of what is happening
in the TCP-IP-implementation in the linux kernel.

This book is simply a rehash of the source code so don't bother spending 60-some dollars when there are better books on the subject. With almost 900 pages, you might think you're going to read a detailed technical presentation of the TCP/IP Linux source much like Wright and Stevens did for BSD. Well, look somewhere else, like the Linux source, because that's all this entire book is with little, if any, added value.

This book was basically the linux source code without the benefits of lxr.linux.no. The book is full of code examples without any explanations. The index is lame and does not include many key terms .

This is one of the worst TCP/IP books I have ever encountered. Originally I had hoped it would be similar to the Stevens books on BSD, but there is very little narrative and no code annotation. If you simply downloaded the current kernel and printed out the source code, you would essentially have a more up-to-date version of this book.

Like most other readers of this book I'm very disappointed. With such prominent authors I was expecting more than a rehash of O/S and IPC concepts with generous reprints of un-annotated source code. Yes, I can read C code and they did say the book was for those well versed in reading C. However, if I wanted to take the time to read and understand the source code myself then I wouldn't have bothered spending $70. The book does at least give a good idea of exactly which pieces of code do what. However, this book is a far cry from Richard Stevens' books (My Hero !, R.I.P.)
At most I would feel comfortable with paying $20 for this book for reference purposes and because of the nice hardback binding.

Overall, the book is pretty pathetic and the authors should be embarrassed to have ther names associated with it. There is still room for a GOOD book about the TCP/IP implementation on Linux.

Anyone in modern network research knows about "TCP/IP Illustrated, Volume 2: The Implementation" (TCPv2) by Richard W. Stevens and Gary R. Wright. That book is a true gem.

By it's description, this new Crowcroft and Phillips book should to the Linux networking system what TCPv2 is to BSD networking. I ordered the Crowcroft and Phillips book immediately when I learned of it, hoping indeed that it would be the definitive treatment of Linux network implementation, based on Jon Crowcroft's distinguished reputation as a networking reasearcher.

Unfortunately, Crowcroft and Phillips has much more in common with previous efforts to illustrate Linux networkinig code than the venerable TCPv2. Unlike TCPv2, the text in this book is not really integrated with the Linux source code at all. There are many sections that are little more than a verbatim dump of Linux source code, it's quite shameless in my opinion.

For people interested in the networking internals of Linux, it is sad that no one with the talent of the late Mr. Stevens has stepped up.

I think the bottom line with this new Crowcroft and Phillips book is that studying Stevens' books together with the Linux source code, is still the best way, by far, to learn about the Linux networking code.

The book cotnains about 1/3 text, 2/3 code.
The book is not meant to be a replacement for, or even
competitor of the amazing Stevens books - we are simply not worthy or able to do that. nor do we need to - luckily for the
internet community, Rich already did.
The book attempts to show where the protocol code glues to
the OS (something radically different in linux than bsd) and to
cover in more depth topics that are novel (e.g. forwarding treatment), rather than internet protocol topics that are well done elsewhere (e.g. TCP/UDP)

The code included was broken down into readable chunks and soem attempt to organise this into topics. Of course, linux is fast moving (2.5 wil be out real soon now) though luckily, much of the networking code was in place in the 2.4 and is less quick to change now ...we intend to update and expand the text via the web, and a new edition, as we see 2.5 come out - we hope.