JMS Client - Server Design and Perfomance Query

Hi all,

I have a question on how to obtain maximum efficiency and perfomance for JMS Client - server application. I have TIBCO EMS as the JMS service provider.

I have multiple clients (separate systems) all having sending similar requests and expecting appropriate responses. My question is how should I implement the solution on server side form perfomance and efficiency point of view of the clients, server and the TIBCO EMS server. I have 10 clients.

Whether I should use a set of separate request response queues for every client or should I use a single set of request response queues and provide the clients with a different message selectors so they can fetch responses from the same queue.

The main criteria is to have minimum response time from the server but at the same time making sure that neither the server application or the TIBCO EMS server is overloaded.
This will be a real time application with quite a heavy load.

Thanks in advance!!!