Hi,
I hope this message from tc-developers list can help you!

Cae

"T. Kandasamy" wrote:
> 
> Hi
> 
> In our organization, we are trying to put the tomcat 3.3 in the production
> environment. I wish to see
> some performance test results of the tomcat. Is it highly stable?. If any
is
> using the tomcat in the
> production/ live server, please share your experinces with me.
> 
> we have successfully integrated apache web server with tomcat. I unable to
> see performance results
> in the jakarata.apache.org. How concurrent requests it will handle?. we
are
> having redhad linux with
> processor and 4gb ram machine. please give your reply

I wouldn't rely on any benchmarks or experiences of others since they
could be a result of a totally different scenario then yours. There is a
simple tool called 'ab' that comes with Apache. It is extremely easy to
simulate server load with it. I suggest you try it in your environment
and make decisions for yourself.

Bojan

PS. My own benchmarks show that a P2 266 with 288 MB of RAM, 2 UW SCSI
disks, running RedHat Linux 7.0 with selective Rawhide updates, kernel
2.4.14, Apache 1.3.22 and TC 3.3 can serve around 60 requests per second
(or 17 ms per request) at concurrency of 10 requests, using one of my
typical Velocity pages... which is probably totally irrelevant for your
scenario.

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



-----Mensagem original-----
De: Jacek Kempski [mailto:[EMAIL PROTECTED]]
Enviada em: quinta-feira, 22 de novembro de 2001 15:53
Para: [EMAIL PROTECTED]
Assunto: Architecture


Hi all,

i have a scalability question.

We build a pretty complex multitier production line control system. 
There are about 500 machines delivering data and 500 PC standing next to 
machines and retrieving processed data (monitoring the equipment) We 
have two presentation layers: one is typical reporting "on demand" - no 
problem with that. The second layer however should be active, that is 
actively notified about events. Usually we would do it with Java 
clients, no browsers. But if browsers.... That would mean that the 
server somehow should be able to push data to the browser. Sure we could 
  refresh and poll data, but 500 clients polling every , say 2 minutes 
is too heavy.
What we would like to try is to place the second tomcat next to the 
"processing tomcat" and fill it's global context with messages coming 
from the processing server through server-server call. Thus we would 
have kinda message board, and the clients would refresh to that second 
"message" server. No reading files, just send an id and a string of 
equipment that alerts.
Still, no change at all to 500 clients polling every 2 minutes.

My question is simple: Is this realistic? Or is this going to explode?
Any other ideas?

Thanks,

jacek


--
To unsubscribe:   <mailto:[EMAIL PROTECTED]>
For additional commands: <mailto:[EMAIL PROTECTED]>
Troubles with the list: <mailto:[EMAIL PROTECTED]>

--
To unsubscribe:   <mailto:[EMAIL PROTECTED]>
For additional commands: <mailto:[EMAIL PROTECTED]>
Troubles with the list: <mailto:[EMAIL PROTECTED]>

Reply via email to