Data replication between Datacenters - Best way? Ideas?

[cmhippy]

vSphere 4.1 U2

IBM SAN at both locations.

IBM System Storage SAN Volume Controller (SVC - which virtualizes all backend storage.  Used to manage LUNs/Hosts)

TPC/Replication - Used to setup sessions (a session will have the host and the disk(s) that should be replicated between sites)

2 VMs - each one in a Datacenter.  Each one has a disk with information which must be replicated to the other VM in the other datacenter.

Both disks/LUNs are of identical size.

Since this is in the virtual world, I created a specific LUN -> Datastore for the disk on each server/datacenter.  The datastore does not hold any other VM files.

When replication is to be initiated, the second VM is shut down beforehand and the TPC/R sequence is started.

In the physical world, this procedure was working fine.  Second server (whichever direction the data was to be replicated) was shut down and LUN was taken away from the Host.  Replication initiated.  If we had to revert to the other server the TPC/R sequence was stopped (Suspend/Recover), and the second server was turned back on with up-to-date data.

I have tried what I thought would be similar to physical but I am running into a couple issues.  I'm either missing something or simply going the wrong way.

I will greatly appreciate it If someone with experice with this can give me a hand or point me in the correct direction.

Thanks

Chris

[TheEsp]

Hi Chris

So your issue s bringing up your DR LUN in a state that is read/writable by your DR ESX cluster ?

Are you using the native IBM replication tools ?

Cheers

David

[cmhippy]

David,

Thanks!

Yes that is correct.  But after a couple additional tests I got it going.

However, I still don't understand is the difference or outcome (since I have not tested the other alternative) between "Keep original signature" and Assign a new signature.  I am using "Assign a new signature" because as I see it the replications in the SAN is block level therefore the resulting copy would be identical to the origin and I don't want ESXi to think there are two identical LUNs.

Correct me if I am wrong.

Anyway, these are the current steps I am taking (to help understand my original question):

Note:  I am not replicating the whole VM, just data on disks between sites.

Destination VM steps:

Disable SQL and other services (in our case)

Remove disk from VM

Shut down VM

Unmap Volumes from VM Hosts

Rescan All (Storage and VMFS)

Prepare SAN disk replication to replicate in the correct direction (we use IBM TPC for Replication in our IBM/Brocade SAN)

---

When we wish to bring up the VM again:

Suspend & Recover (makes disks fully available) (TPC for Replication steps)

Map Volumes back to all VM Hosts

Rescan All (Storage and VMFS)

Add Datastore back to Hosts -> ASSIGN A NEW SIGNATURE

Rename Datastore back to original name

Power on VM

Add disks back to VM

A couple more steps than in the physical world but it is working.  Just wondering if I can our should use "Keep original signature".

Thanks to All!