----------------------------------------------------------------
BEFORE YOU POST, search the faq at <http://java.apache.org/faq/>
WHEN YOU POST, include all relevant version numbers, log files,
and configuration files. Don't make us guess your problem!!!
----------------------------------------------------------------
I have a single box configuration that uses Apache
1.3.9, JDK 1.1.8(blackdown), Jserv 1.1.1, LAN(10/100)
Mixed. Under load we have been able to sustain 200
concurrent users with an average 1600 pg/s.
We are running about 85% to 95% of available memory
and cpu. The hardware is a low end Pentium class
machine. We have multi-box configurations that have
held 5000 concurrent users. Ideally you need to look
at page rendition time. A good benchmark is between
200ms-600ms.. If it takes longer than 1 second then
you may need to look at your algorithm or change your
performance expectations.
Memory footprint will be dictated by the application.
Cpu load is dictated by kernel scheduling and number
of processes(Apache & JDK) running on the box. You
also need to watch the I/O through put of the box.
Make sure you are not waiting on the disk drives to
read/write data. There are many other issues to
achieve maximum performance in a setup like this, but
these are the key areas. There typically bring the
most bang for the buck.
David Miller
Megabyte Info. Svcs.
> >
> > I am also very interested in benchmarks.
> >
> > I recently run the following test:
> > - Apache 1.3 on Solaris 2.5.1, dual processor SUN
> Ultra-60
> > - Jserv 1.1.1 with JDK 1.1.6 (native threads),
> same machine as Apache
> > - 10 Clients from different machines in the same
> LAN that perform http
> > requests to the Hello example servlet in a large
> loop.
> >
> > In different test scenarios (less and more
> clients), I always come up at
> > around 100 requests per second as the maximum,
> which can be served. Is
> > this an expected number? Can I increase that
> number without increasing
> > the machines that are running Apache and JServ
> (example: different Java
> > version, starting of more JServs)?
> >
> for my test my config is :
> * one Netra (Ultra II 440Mhz / 256Mb ) with apache
> 1.3.12
> * one E3500 8cpu 3Gb with JServ 1.1 using Java2 SUn
> hotspot.
> * tests clients on the same LAN (100Mb/s)
>
> to handle my test scenario (I have used LoadRunner
> from Mercury and
> generate a test case for my application, there is
> more or less 5 "big"
> requests ie: 1 html servlet + some images.
> at near 300 mercury users my Netra was "out of RAM"
> but pretty cool in
> CPU. Note that I have put 300 apache processes to
> handle this which make
> me think it give something like 300req/s .
>
> -------
> Ouch - that sounds pretty low to me...although the
> judgement really depends
> on what kinds of operations that servlet does.
> Running out of RAM is also not good news :(
>
> Does Tomcat perform better, does anyone know?
>
> Otis
>
>
>
>
>
=====
David Miller
Megabyte Information Services, Inc.
__________________________________________________
Do You Yahoo!?
Send instant messages with Yahoo! Messenger.
http://im.yahoo.com/
--
--------------------------------------------------------------
Please read the FAQ! <http://java.apache.org/faq/>
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Archives and Other: <http://java.apache.org/main/mail.html>
Problems?: [EMAIL PROTECTED]
