This book gets very high ratings on both amazon.co.uk and amazon.com. I have given it a slightly lower rating (although still 4 stars) than most and will explain why below.
The subtitle on the cover of the book is "Practical Guide for Programmers" which suggests it is going to be good even for experienced developers. It is only when you read the preface (page X) that you find that the book is aimed "primarily at students", and even then is "intended as a supplement, to be used with a traditional textbook", which seems a bit of a contradiction when it then says that "we have tried to make the book reasonably self-contained".
Anyway, what are the good points of this book? Well, it does mention most of the bits that a developer using sockets will want to consider. It has everything from blocking sockets, through non-blocking sockets and the select model, through to overlapped I/O. It also mentions threading, the use of thread pools, broadcast and multicast. All good stuff. Even includes example code for each.
Where the book falls down is that having skimmed over all of those topics it (a) doesn't provide adequate information about how to choose the model (synch vs. asynch, blocking vs. non-blocking, 1 thread vs. fixed number (> 1) of threads vs. thread pool, etc) to use for a particular project, and (b) falls short of being self-contained, doing the blah-blah is beyond the scope of this book thing.
I have seen many projects developed using the wrong model, resulting in poor performance, lack of responsiveness, inability to shutdown cleanly etc. I'm pretty sure that the authors of the book will have seen projects like that too. Books about using sockets really need to advise on this area.
It is understandable that a book of this size and price will say that some things are outside the scope of the book, but not something as basic as socket options (p52 refers the reader to the MSDN). Again, socket options are an area where well-meaning developers or support staff set values that are little better than guesses, and which sometimes cause adverse effects. If there's going to be a second edition of this book, please include advice on such matters.
So, all in all, good for students or people new to sockets, but not quite great. It tells you the basic techniques, but not how to use them to best advantage. Having said that, I prefer this book to C# Network Programming which rambles, uses language that is ambiguous in places, and contains a significant error (if being very generous, it could be very lazy English causing an unintended meaning) on the very page I opened it on.
It's probably best for people who already know sockets really well, but who are switching from one language to another (e.g. C++ or Java to C#). Those people probably know what model and options to use, just need to see how to do it in C# - something the book does do well.