Clustering in a distributed system

[andrew.henderson]

I am trying to determine if JBoss can help me in my current situation. I have servers in two different geographic locations linked by a corporate WAN. I would like both application servers to share configuration information and configuration data, but I would like to regulate which node is being used so that all requests from location A go to location A's node and location B's requests go to the node at location B. The objective behind this is two fold, first the performance will be higher if not bogged down by the WAN and second, it is preferable not to send the data that is actually being passed, processed and returned across the WAN for security reasons.

Can my situation be resolved using Jboss clustering or some other feature of JBoss?

[justkeys]

When a client uses jndi to reach a clustered jboss service or ejb, the Context.PROVIDER_URL property of the jndi context refers to a comma-separated list of hosts in the cluster. So if you want clients on some site to see only host A, then you can pass only node A in there.

Or if host B should still be used for these clients when node A is down, then set the property to A:1100,B:1100. And change the order for clients on the other site.