my tests are heavy on JSTL tags, so that might negate some of the
performance enhancements.  With 4.1.3 I did notice JSP pages with
scriptlets considerably faster than 4.0.3. Overall response time for
static and scriptlet pages were faster than 4.0.3.

if I have time, I will run my scriptlet tests with 4.1.7.

peter



Remy Maucherat wrote:
> 
> peter lin wrote:
> > I'm running 4.1.7 now using my test webapp.  I don't see a huge
> > performance improvement over tomcat 4.0.3 with jasper2, but it does look
> > more stable and reliable.
> 
> I still think that it should be quite a bit faster, but it depends on
> what your benchmark is measuring.
> 
> > the CPU profile and response variance are considerably lower. The
> > standard deviation (read variance) using Jmeter is 1/2-1/3 that of
> > tomcat 4.0.3 with jasper2.  Most of the response times are in a narrower
> > range and wild variance is much less.
> 
> That looks good. Coyote lowers GC considerably, and will perform a lot
> better with real-world requests than the old connector.
> 
> IMO, 4.1.7 should be the most stable and reliable release of the 4.x
> branch so far. Of course, we'll see how good it actually is after
> extensive testing ;-)
> 
> Remy
> 
> --
> To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to