Just wanted to get some idea from the more seasoned architects here about my customer's situation.
They need to ensure there's no downtime for their Java application running on top of JBoss EAP configured as a cluster, VMWare FT enabled.
My noob question is, do I still have to configure RHEL Cluster suite for this scenario?
I imagine that since FT involves the VM being synced(full mirror) between a set of hosts, I wouldn't need RHCS anymore (or do I?).
2-node VMWare vSphere cluster (Fault Tolerance enabled)
2x RHEL VMs
2x RHJBoss EAP (one instance per VM)
Would really appreciate your feedback! :)
Without more information about you particular use case, I would use JBoss EAP native clustering features. They provide an active-active cluster of many JBoss EAP instances, and as far as your applicaitons follow recommended practices for Java EE development, they should 'just work'.
You do not need any shared storage, nor any cluster middleware such as the RHEL Cluster Suite to provide hight availability for JBoss EAP instances and other Red Hat middleware deployed on top of JBoss EAP.
AFAIK any solution based on shared storage from virtualization providers and also the RHEL Cluster Suite are active/passive solutions, so they incurr on some downtime while the virtual IP migrate and storage is reattached on the surviving node. That downtime might be very small but it is measurable. Also remember that these solutions requie that you start the passive JBoss EAP instance.
Solutions like JBoss EAP clustering, while active/active, may fail to reply to a few requests while user sessions fail over to the surviving instance. So there is some downtime, but way just less than with active/passive clustering solutions. There's no such thing as "no downtime" in the real world. Just "acceptable small downtime".