NetApp Backups from Storage Snapshots best practice

[ncolt]

Hi, we've just bought an Enterprise Plus license and looking to use the NetApp plugin to enable hardware snapshots.

We only have 1 volume per site of several TB and around 25 jobs per site (which run from 17.30 to 10am the next day). Hopefully when I enable this feature , it won't try to snapshot the multiple Terabyte volume each time a new job starts? How does it decide when to destroy the volume snapshot? Or have I misread how it works.

I'd just like to get a better idea about what to expect when I enable this feature and if our environment needs to be changed to accommodate.

[Vitaliy S.]

Hi Neil,

You have understood it correctly, every job will trigger its own storage snapshot. Please see a step-by-step description on how it works. The backup/replica job:

1. Analyzes which VMs in the job have disks on supported storage.
2. Triggers a vSphere snapshot for all VMs located on the same storage volume. (As a part of a vSphere snapshot, Veeam’s application-aware processing of each VM is performed normally.)
3. Triggers a snapshot of said storage volume once all VM snapshots have been created.
4. Retrieves the CBT information for VM snapshots created on step 2.
5. Immediately triggers the removal of the vSphere snapshots on the production VMs.
6. Mounts the storage snapshot to one of the backup proxies connected into the storage fabric.
7. Reads new and changed virtual disk data blocks directly from the storage snapshot and transports them to the backup repository or replica VM.
8. Triggers the removal storage snapshot once all VMs have been backed up.

Thanks!

[ncolt]

OK thanks. This sounded crazy thinking about the way VMware does snapshots but the storage guy has reassured me that the way NetApp takes snapshots, this method should be fine for our setup.

[Vitaliy S.]

Yes, you should not see big impact on storage and especially on VMs when using storage snapshot integration.

[Gostev]

NetApp snasphots are extremely efficient and carry virtually no I/O overhead.