Hi,
Welcome to the headache that is RDM!
I will be speaking from experience of RDM via 6.0 (6.5 has some new methods that may have remedied my arguments).
1. Cloning? Nope! Snapshots and (host level) backups go out the window too.
2. vMotion? Yes, as long as all the ESXi hosts have the appropriate paths.
3. Storage vMotion? Only the pointer files will be moved, not the actual data.
4. Fault Tolerance? Of course, I think you can setup some rules to keep it from failing over to the same host. Replication fail over would be a completely different story.
Two more annoying caveats:
1. You will see ridiculous boot times unless you tell each ESXi host to ignore the RDM disk (about 15 minutes per RDM)
2. It is not straight forward in any proper sense. And anyone casually working on your vCenter (such as VMware support or the FNG) can really screw something up because there is typically nothing that screams RDM. KEEP GOOD NOTES!
I am at a cross-roads of rebuilding the RDM mappings with 6.5 or flipping them back to iSCSI. I have been using them for a Windows Storage Cluster and have seen no real benefit. Maybe someone with a SQL cluster can chime in?
