VM consistency with SRM and IBM DS Series

[VmwareProfile]

Hello guys,

How ensure VM consistency between 2 LUN with IBM Products in case of total sinisters of a protected site?

SRM is in charge with SRA to coordonate the failover by mounting appropriate volume, but it seems there's nothing about consistency check.

I know for NETAPP, in asynchronous mode, a tool called SnapManager communicate with vCenter to snapshots VM and launch a block replication to ensure consistancy.

Thank you for your help !!

[admin]

Hi,

If you are asking about the ability to restore to a specific virtual machine snapshot,  SRM allows for a hook that array vendors can use to implement such a feature. Please consult with your storage vendor if they use / support this feature.

Thanks

[VmwareProfile]

Hi,

No, my question is how to be sure a vm is consistant when she's replicated (synchronous ou asynchronous) by system storage from site A to site B ?

Thanks

[admin]

I don't know enough about IBM DS series to answer for every replica. However, for testFailover or disaster recovery, SRM provides enough flexibility for the storage vendor to implement reverting back to a VM at consistent snapshot state.

-Sudarsan

[VmwareProfile]

Hi,

In your response I don't understand how storage vendor can reverting to a consistent snapshot. It's suppose the snapshot processing and the replication processing are coordonates, so who is in charge to coordonate these 2 tasks ? SRM (by a SRA) or a specific tiers vendor sofware?

I understood SRM doesn't manage the replication but just the failover process.

An other point, if snapshot processing and the replication processing must be coordonates, it's suppose a synchronous replication is impossible due to vm snapshot time execution? True ?

Thanks to light my mind !

[admin]

Hi,

In brief, here is how it works (for async replication only):

1. SRA is capable of communicating with another server (let us call it TP) which can take consistent snapshots of a VM (I think some vendors use in-guest agents that can help with application consistency too). TP does this by communicating with virtual center.

2. For every sync, TP and storage array work together to make sure that the replica has consistent snapshots of VMs. Same thing happens for the sync request initiated by SRM when user runs testFailover.

3. SRA tells SRM the name of such snapshots.

4. When SRM fails over (either real or test), after bringing up datastores on the recovery side, SRM looks for snapshots with the name given in #3. If SRM finds such a snapshot for a VM, the VM is reverted to this snapshot.

I don't know if such a thing is possible for synchronous replication.

Hope that helps.