HPE Storage Users Group

Hi All;

I could prob. log a call with the vendor but thought I'd get quicker response on here.

So we are migrating our large ESXi cluster storage at the weekend, T400 -> 7400
using peer motion, this has worked well for us on smaller clusters, this current size
is around 60TB.

We do have a little concern about the Commvault 'dedupe' raw data from the C-tree
database, this sits on 3PAR LUNs in the ESXi storage cluster. Obviously we would freeze
backups but no sure how the migration would be treated.

Anyone tried this migration using peer motion ?

Thanks for any speedy info.

Gary...

Why are you worried? Peer Motion Spoofs the new LUN so it looks like the old LUN. My issue with Peer Motion in general is the spoofing that occurs, my understanding it the new volume has a physically different WWN/VOLID and the old info is spoofed out to the hosts. I am not a big fan of spoofing/emulation.

Hi,

Not worried, just concerned as we have used peer motion on oracle and sybase DB's
which used a filesystem but not a LUN with raw data ....

Gary.

Certainly current Commvault DDB is stored in files, on a NTFS disk in our case. We've stopped services and copied the files between physical disks several times. Commvault only tends to care about drive letters/paths for both the DDB and block datastores. I assume you are using a RDM disk presented to a Windows VM running on ESXi? I'm not sure how Peer Motion handles RDMs but I presume it's doing all the copies at a block level, I wouldn't imagine a problem. If worried just mount both disks in the VM, copy the file structure across, do a binary file compare to be sure, swap the drive letters over and restart Commvault.

To my knowledge the only major problem with peer motion currently is persistent SCSI reservations in windows. However according to HP that is changing in 3.1.3.

_________________
Nathan Bell
Solution Architect Principal
Dyncorp International