Limit number of simultaneous executing calls.

[danielen]

We need to limit the peek throughput so that we dont DDOS all of our sub systems (db, external systems etc) that has a limit of how many clients calls they may execute at the same time.

We are migrating a weblogic application to jboss. Weblogic has execution queues that limit how many simultaneous rmi executions that could happen from all connected clients, and make sure that we always are safe within of the throughput limits of the system.

The only solution we have found right now,
pray that we never have any peaks in rmi trafic and that we can handle the normal application load within limits, is not good enough for production.

This is a EJB2 system with Stateless beans, using jboss 4.3 eap.
If no solutions in the ejb2 container, does the new ejb3 container support somthing we could use?

[dmlloyd]

What policy would you impose when there are too many concurrent requests? Block/enqueue the request? Reject it outright?

[danielen]

We would like to enqueue rmi calls on the server side to handle peak load which could happen in short periods.

[luc.texier]

Hi danielen,

Could you please send me an email directly so we can discuss the best course of action in order to address your issue in a timely fashion?

Cheers,

Luc Texier
JBoss Support Team
ltexier @ redhat.com

[danielen]

this topic is related to filed support case: https://support.redhat.com/jbossnetwork/restricted/caseDetail.html?caseId=245005