Consistency Groups

[ohuetz]

Hi,

does anybody know how to create some kind of "consistency group"?
We have several server who have a relationship (database / application etc.) and should be backed up at the same moment. Best way would look like
1) Shut down all servers of a group
2) Take a snapshot
3) Start all servers of a group
4) Backup the snapshot
We want to keep the downtime as short as possible and can´t wait until the backup is done to start the server again. An alternative could be to stop all related services instead of shutting down the server. Most important point is to take the snapshots at the same time.

Does anybody have an idea?

Kind Regards
Oliver

[foggy]

Oliver, have you considered using backup from storage snapshots? This should address your requirements.

[dellock6]

If you move to Veeam v7 you can leverage parallel processing for doing this.
You would need to have enough processing slots as the number of vmdks of the "consistency group"; let's say you have 3 VMs with 2 vmdk each, if you have at least 6 processing slots, with parallel processing all those vmdk will be snapshotted and saved at the same time (obviously if there is no other job already in the queue occupying some slots).

Luca.

[ohuetz]

@ foggy
Yes, exactly. But we use IBM Storage and so I have to wait for support.
Perhaps someone has an idea to solve the problem with tools or skripts which exist now (except VEEAM tells me that they will publish IBM Storage support in the next month )

@dellock6
I already thought about this scenario, which could be realized with a couple of proxies, but is there a way to start the VMs right after taking the snapshot (not all databases support vss)? I could shutdown the VMs as pre-backup-job but I can´t find a way to start them at the right moment (post-backup-job will be to late).

[Gostev]

I suggest stopping/starting application services with pre-freeze and post-thaw VMware Tools scripts instead. This is waaay faster than shutdown/startup cycle

[ohuetz]

OK, so if I put all together the best solution for the moment will be to create a separate job per group with parallel processing und use pre-freeze and post-thaw scripts to stop the applications. I will check if that is a possibility because there are several groups and the backup-window is short for all together. I will check storage-snapshots as soon they´re available for IBM.
Thanks for all suggestions. If anybody has more ideas, please let me know.

[dellock6]

Oliver, just to be sure each CG is processed completely, schedule each of them separated in time, so they enter the queue with different priority and each one is processed from start to end without overlapping.

About the IBM storage, which one are you referring to? V7000?

Luca.

[ohuetz]

I'm referring to V3700.