I have done a day or two of profiling with OptimizeIt.
We're spending about 20% of the time pulling code from .zip files. This is probably a JBoss/Tomcat problem. It looks like this is related to the problem where Tomcat-4.0 (embedded in JBoss-2.4.3) does NOT read its initialization file (server.xml).
It doesn't look like the Tomcat half pulls .class files out of .zip files into "work" directory. TC's "work" directory does get the .jsp files but it doesn't look like the servlets get pulled from the .jar file that's buried inside the .ear file. Tomcat (inside JBoss) doesn't even un-zip all its .jar files.
Does anyone know if the JB/TC folks are fixing these problems in the upcoming JBoss-2.4.4_Tomcat-4.0.1 release?