Just added two new connection strategies for the clustering/failover logic.

RoundRobin: As the name implies each request just goes to the next server in the list. It does this on every request as sort of a cheap load balancer.

Random: Similar to the above but it picks a server randomly from the available servers, excluding the previously invoked server if possible. So you should hit a new server, but not necessarily rolling down the list in order.

I'd like to create some way to configure this for an individual app (or even an individual bean). Thinking maybe some annotation and openejb-jar.xml meta data that we can swizzle all the way to the client code.

Thoughts?

-David

Reply via email to