Running RMAN for Hourly Backups vs Storage ARRAY LUN Snaps
593025May 23 2012 — edited May 24 2012Hello all,
I have a 2-node oracle 11.2 cluster with storage using ASM. The two nodes are both VM's and their OS-level ASM disks (2) are independent disks shared through their hypervisors.
We currently protect the oracle databases by running full RMAN backups nightly, but I'm looking for a way to offer hourly protection in case of data corruption. Our iSCSI storage array performs hourly snapshots the LUNs that hold the VM's and the ASM disks (they're in separate LUNs). We don't snapshot the VM's through Vmware because they can't sit still long enough quiesce and the VM's hang when we try.
My question is:
Is it better/more efficient from a DR standpoint to stick with the LUN level snaps for hourly protection? I figure the best DR procedure would be to simply shutdown the nodes, mount the snapshot in place of the corrupted LUN, then boot my DB node VM's and cross my fingers. Or, perform hourly RMAN online log backups and deal with their storage, then perform an RMAN recovery using the last full backup and the trailing incremental ones.
What possible pitfalls am I looking at if I choose to rely on the array snapshots? Is there a better third way?
Thanks,
Dan
Edited for spelling by: ddenton on May 23, 2012 12:45 PM