Title: JSP Server Performance
 
I would love to see test between the different JSP and Servlet engines. But please, no more informal test. There are people out there that make haste decision based on stuff like this.
 
There are just too many unknowns in your test:
 - What versions of the engines did you use.
 - How much memory does your computer have.
 - Which engine got loaded first (e.g. hogging RAM), did you try starting them the other way around.
 - What happened when you ran the test one engine at a time.
 - Which settings did you use
 - What web server configuration did you use for tomcat, apache or the java server.
 - How did the engines scale to bigger JSP pages than hello.jsp.
 - Did you find any reasons for the dropped requests?
 
Anyways not much work has been done to optimise Jakarta, yet. There are other priorities right now (plugable internal architechture and other cool stuff). It would probably be more fair to use the next release.
 
/Antonio
-----Original Message-----
From: Griggs, James [mailto:[EMAIL PROTECTED]]
Sent: den 8 februari 2000 16:30
To: [EMAIL PROTECTED]
Subject: JSP Server Performance

I ran an informal test for 9 hours last night on a Win98 PC comparing the latest tomcat(apache.org) code and the latest resin

(caucho.com) both using Suns JDK 1.2.  I used 2 instances of socrates to simulate 100 users requesting a hello.jsp file at random intervals.    Tomcat and Resin were running at the same time on different ports.

Result:Tomcat:  Resin:
Total Requests Handled          260946546375
Total Requests Dropped          4691          0

Requests per Second         .1              195

Average Latency98.17      .15

I'm assuming the dramatic difference in requests handled is due to a 60 second timeout on dropped requests.

I'm wondering if anyone can explain the dramatic difference in the results?  Am I doing something wrong here to cripple Tomcat?  (... and since I know the question will be asked: I have no affiliation with Caucho ;o)

Jim Griggs
[EMAIL PROTECTED]

Reply via email to