I’ve been a loyal Resin user for about 3 years now. For a time, it was one of the fastest servlet containers. Well, I have an web-app, made up of 3 wars. It uses a number of open-source, 3rd party libraries, such as Hibernate, Spring, Acegi, DisplayTag. The length of startup time has never really been an issue. However, after upgrading from resin-3.0.19 to resin-3.0.23, and deploying the expanded war, startup time went from 90 to 117 seconds.
Since I simply deployed the expanded war from on instance to another, I knew it had nothing to do w/ the war contents. Rather Resin was doing something different. The majority of load time seems to be during the precompiling of jsps. I wouldn’t mind the extra time, if a benefit was visible at request-time, but I didn’t see that. Although just a visual inspection, the pages returned in about the same speed.
I was initially writing to see if maybe i was imagining this — and then I saw a comparison done by Matt Raible. Guess should to take a closer look at Tomcat-5.5.
Go for Tomcat 6.x instead of 5.5 if you don’t have application hosting limitation. It’s faster both on startup and on runtime (using NIO)…
Thanks, I’ll check it out.
Pingback: that’s great… » resin-3.1 performance