Without transmission control protocol/Internet protocol (TCP/IP), the Internet would not exist in its present form. Overall, one can say that TCP/IP was invented with the Berkeley socket layer. So, without exaggeration, it can be said that operating systems and TCP/IP are the base of modern computing. There are just a few books available in this area, and Stevens’ [1] is one of the most prominent. However, as far as I can tell, there is no other book about the implementation of TCP/IP for Linux. So, in this regard, this book is unique.

The book consists of 20 chapters. I will not discuss them in their entirety, but, instead, I will separate the chapters into a few themes. The first few chapters introduce TCP, followed by the base Berkeley software distribution (BSD) sockets implementation and how the TCP kernel structures are implemented. The next chapters explain how one of the main data structures runs through the layers, and how every layer adds a bit more information to it. Then, the actual sending and receiving of packages is discussed. A more difficult part follows: memory, collision, and segmentation handling. Next, the authors describe the diverse states of packages. They show the lower levels of the protocol, beginning with the flow of data between kernel and user space. The introduction to the IP part follows--routing, quality of service (QoS) filtering, and the low-level handling of data in the kernel. The book ends with a chapter on TCP/IP debugging.

Regarding readership, according to the authors, the book is targeted at researchers, academia, corporations, entrepreneurs, open-source users, and defense organizations. I disagree with corporations and entrepreneurs, as these groups are interested in the application of services and providing applications to users; therefore, I don’t see how this book would be of interest to them. I would, however, extend the readership to people who are interested in the real-world application of C. C is not an academic language and it’s not that interesting for research, but it is the workhorse of the whole infrastructure. Its largest part is network programming. For these people, while the book will be tough reading, they will see the extent to which C can be used.

Now we come to the weakest element of this book. The book was written for Linux Kernel 2.4--I do not understand why. Kernel 2.6 has been around since 2001; I checked the sources and one can find differences between versions in one of the most central data structures used. If the audience is indeed academics and researchers, they are forced to interpolate. The base data flow may be recognizable, but what about the fields that are not part of the 2.4 data structures? The authors only mention Kernel 2.6 on three pages, at the end of the book. Another area completely missing is the implementation of IP version 6 (IPv6), in which people are increasingly interested.

Another weak point is the index, which is just four pages long. There is no glossary, so readers must write down the abbreviations they encounter. The order of the presentation is mostly good, with a few exceptions where code is discussed and printed ten or so pages after the point where it is mentioned in the text. The proofreading is good, not perfect.

Overall, I’m disappointed. The book is not current and omits many parts. However, because it’s the only book of its kind, those interested in the subject can use it as a starting point for their own research. The best way to use it is to be very selective. The book contains some overall dataflow diagrams that will guide readers. While this book will assist you, it does not hold up to the promise of the publisher’s site: “This book provides thorough knowledge of [the] Linux TCP/IP stack and [the] kernel framework for its network stack.”