Has anybody seen any actual performance numbers comparing Tomcat
5.5.x/JNI/APR with just plain Tomcat 5.5.x? Having built the shared
libraries and installed them into my JRE, about all I can say is that using
the APR makes configuration harder. The Tomcat doc makes noises to the
effect that it "should" be faster in "some circumstances", but I can't find
any actual benchmarks anywhere to confirm or deny up that particular
statement. At the moment, my inclination is to just stick with straight java
Tomcat on the grounds that it's one less moving part, but I'd like to get a
feel for how much of a performance hit I'd actually be taking if I did so.
Does anybody know of any testing to that effect?
--- Pat
PS In case you haven't looked into this yet, recent versions of
tomcat let you use JNI to natively invoke the (c language) apache portable
runtime library on your server. So instead of having Apache fronting tomcat,
you have tomcat calling through into the Apache Library. In theory this is
easier to configure than apache->tomcat, but still faster than just plain
tomcat. In practice, well, that's what I'm asking about.
