Hi Plug Bert,
#Can anyone give me a list of web servers with JSP
#support? Your comments/suggestions as to which is the
#best would be welcome. Thanks!
#
You don't want to use a standalone JSP engine to be both the JSP engine,
and the webserver -- webservers such as Apache can do a lot better in
serving up static files than any other webserver. It's for your good
interest to make it that way :)
I've used Tomcat, Resin JSP, JRun, and Weblogic Server (WLS) as JSP
engines fronted by one or more Apache instances (using supplied
Apache module). I've also load tested each with 1000 concurrent vusers
using a load testing software called LoadRunner (by Mercury
Interactive). As a base line test, I had a single Apache instance in one
box, and a JSP engine on another. The test was on our web app -- JSPs
(accessing JavaBeans, and EJBs), and static files (.jpg, .html, .js,
etc.).
The following are the results (I can't publish the official
numbers.. they're proprietary)
1. Apache->Tomcat: Performed horribly. Tomcat chocked, and dumped at a
little over 100 users. After cleaning up the configuration file, and
removing some unnecessary services, I was able to get Tomcat to work with
close to 300 sessions. But that was it.
2. Apache->JRun: It performed well under load. But I couldn't get it to
go more then 400+ users. I it left at that.
3. Apache->WLS: After a few painful experiences configuring the Apache
module of WLS, I was able to get it to properly work. The problem
was caused by a bug in the WLS Apache module -- it didn't allow JSP-only
request-forwading to the WLS instance.
With only one Apache server fronting the JSP engine, I was only able to
get to 600+ users. After that many users, the Apache server locks up the
new users, and the CPU utulization was already 95-100%. So to get to
1000 users, I had to setup another Apache instance in another box, but
still use a single WLS instance. After that, I got 1000 users happily
clogging-up my network. ;)
4. Apache->Resin JSP: Of JSP engines I've tested, Resin had the best
performace. Is had the best throughput for JSP processing -- as per
throughput computations of LoadRunner.
Though I had to also add another Apache instance (as with the WLS setup),
I was able to get to 1000 simultaneous users. Even at 1000 simulteneous
users, CPU utilization was hovering between 70% to 90%.
In conclusion, I would like to stress out to NEVER deploy a JSP engine as
both JSP engine, and web server. They won't perform as much as if they
were used only as a JSP engine.
If you have any questions, just post them here at the PLUG mailing
list. BTW the machines were single CPU-Netra T1s with each having 500MB
RAM. All were connected to a 100baseT switch.
stay cool.
jeff --
--
Jeff Gutierrez
Mapua Online!
http://www.mapua.org
http://www.mapua.com
http://www.mapua.net
Pinoy Ako! May reklamo?
_
Philippine Linux Users Group. Web site and archives at http://plug.linux.org.ph
To leave: send "unsubscribe" in the body to [EMAIL PROTECTED]
To subscribe to the Linux Newbies' List: send "subscribe" in the body to
[EMAIL PROTECTED]
