I have a couple of problems understanding how to integrate EclipseLink with JBoss from a conceptual point of view. As JBoss usually comes with Hibernate and both of them use JBoss Cache and JGroups I would assume that beans and entities are in sync on all cluster members.
With EclipseLink, you use RMI or JMS for synchronizing cluster member caches. This is what they call "coordinated cache". And this is the point
that I dont really get.
As there are 2 separate ways of synchronizing the cluster members. All stuff that comes from/to the JBoss servers use the Jboss Cache and the underlying JGroups and the JPA caches use RMI/JMS. How does the cluster guarantee to be in sync all the time. Is there a chance that you have stateful beans that dont reflect the current status of their backing entities as they are not yet synchronized? Or does the whole thing attach to JTA and there are transactions that guarantees all members to be in sync?
I am pretty new to all this stuff, so chances are pretty good that there is just a lack of knowlege. So I would be happy if someone could enlighten me.