This week's book giveaway is in the OCPJP forum. We're giving away four copies of OCA/OCP Java SE 7 Programmer I & II Study Guide and have Kathy Sierra & Bert Bates on-line! See this thread for details.
I was reading a response from one of the authors in another thread (I went back to find it to reference, but I lost it in the pile), the author said this book focuses on the administration and performance aspects of Tomcat 6.0.
Did you (in the book or elsewhere) try to quantify the performance gains with any benchmarks? I'm curious how much of the improved performance is attributable to Java 6 (I've heard good things there), and how much credit goes to Tomcat. I'm VERY curious to how much it has improved, even though your benchmarks might not be an accurate representation of our systems.
I posted earlier about clustering. Management at my company loves the "Projected versus actual" performance gain charts, even when actual is below expectations. I'm keen to see what the book has in this area.
Performance is highly sensitive to the nature of the application, the environment (hardware/OS), work-load, and actual configuration.
Generalized "percentage performance gain over XXXX" claims are meaningless, and only useful in marketing material for competing commercial server products.
We try to get this point across in the book.
One great thing about open source servers, such as Tomcat 6, is the fact that you can benchmark your own application on them, running on your target hardware/software environment, using your own custom configuration... at no cost.
The book provides information on things that you can configure and tune, that may have an impact on performance for your specific situation.
Joined: Jan 18, 2005
Generalized "percentage performance gain over XXXX" claims are meaningless
I would have to respectfully disagree. When all other factors are constant (or as constant as possible), and just one is changed and the results measured, it is useful information.
Claims of performance gains? Well, how much? 10% in benchmark tests? .000002% in benchmark tests?
Of course we'd have to benchmark it ourselves with our own application to see what we'd get, and our results will certainly vary. Management isn't going to be interested in spending the time to do this if the expectations are so low from the start.