This book attempts to define a method for addressing performance problems with software, with some particular emphasis on spotting problems before they manifest themselves.
The authors have a distinct method for handling performance issues within the development cycle. They emphasise the importance of tackling performance items as early as possible in the development cycle (taking a few sideswipes at XP on the way).
The method generally seemed sensible, although there were no eye-opening insights and the book seems short on guidance of how to sensibly make the requisite estimates to drive the process.
The book had a couple of annoyances. For example, the chapter on 'Web Applications and Distributed Solutions' had entire paragraphs quoted from the chapter on 'Web Applications'.
On the plus side, the presentation was good and the book was quite readable.
There were some useful ideas in the book. However, I felt they were somewhat few and far between and although they were an interesting summary, they were not a revelation.
The main problem that this book has is that it falls between two stools. Although some of the ideas in the book might be interesting and useful for someone with little experience in performance management, the book is somewhat specialised (and long) for that audience.
It might form a useful chapter in a more general-purpose textbook. For a performance specialist, I didn't feel that the book offered enough of interest to be worth buying.