Controlling load on a a machine with multiple tomcat instances

[Donie Kelly <do...@tecnomen.ie>]

Hi all
We are deploying multiple applications on single machine using multiple
tomcat instances. There are multiple machines with the same copy of each
application on each machine. The whole cluster is hardware load balanced.

The problem is the individual machines are not that powerful (Sun T1's) and
each application starts it's own thread to handle the request because it
take a long time to process. This way we are able to give a fast responses
back to the client while the request is actually still being processed.

However, this brings up the problem of what happens when the machine is
maxed out with all the background processing and lots of new requests
arrive. We want to ensure the VM does not die due to overload.

We are looking at using a CPU load average to decide whether we can start a
thread or not but I'm not sure how successful this can be. Has anybody got
any experience of controlling loads like this or point me at some
applications or code that can help in the design.

Regards
Donie 


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>