Does Veeam snapshot volumes that are excluded from a backup?

[bparson]

I have a VM that serves as a NAS server. The VM is running Windows Server 2008 R2. There are 2 disks attached to the VM. One disk is for the O/S (C:), resides in datastore (DS1) and is sized at 60GB. The other disk is for storing files (D:), resides in another datastore (NAS) and is sized at 2TB. Furthermore, the .vmdk file for D: is sized at 1.96TB (Thick Provisioned Lazy-Zeroed). This disk is NOT Independent.

While we are awaiting approval to purchase more storage to backup D: I excluded this volume from being backed up via Veeam. I ran this backup job for the first time this weekend and I suspect that Veeam may have taken a snapshot of the D: volume even though it was excluded. This is because 1) the job never finished for some strange reason and 2) the volume filled up with several HUGE .vmdk snapshot files.

I would like to know if Veeam indeed does snapshot disks that are excluded from backups?

[Gostev]

Veeam itself does not implement VM snapshots - those are handled by VMware. And the way VMware snapshots are designed, is exactly as you describe (all VM disks are affected). However, Independent disks functionality is there specifically for you to be able exclude virtual disks from a VM snapshot. So, if you want this specific volume not to be included in a snapshot, then you should just make it independent.

[bparson]

Thanks for the confirmation.

[cby]

Be careful when specifying independent disks. I have found that migrating an independent disk to a new datastore is prevented when the VM is powered on. I seem to recall that I had to power off the VM and make the disk non-independent in some cases.

[Menton]

Hello.
I have just installed Veeam 6.1 for the first time to backup my vSphere servers from my SAN to my Exagrid de-dup cluster.

Im looing to backup only the c:\ drives on the servers so im going to use the "Exclusion" function in the backup wizard to only backup SCSI (0:0).
vSphere to my knowledge will snaphot the entire vm with all disks that are not flagged "independent".

Will Veeam be able to exclude disks that are "dependent", that is to remove the disks from the full snapshot?
Or can I only exclude "independent" disk?

I know about the recalc bug som Im not worried about the "total size estimate" .
I just want to know how Veeam reacts to "dependent" disks and exclusion.

Best Regards
Sverre

[Vitaliy S.]

Hi Sverre,

vSphere to my knowledge will snaphot the entire vm with all disks that are not flagged "independent".

Yes, that's correct.

Will Veeam be able to exclude disks that are "dependent", that is to remove the disks from the full snapshot?
Or can I only exclude "independent" disk?

Only independent disks can be excluded from snapshotting. If you set disk exclusions in the Veeam backup job, then VMware will still create the snapshot for the entire VM, but the backup job will only backup the drives you've selected (system drives).

Thanks!

[Wlacroix]

I think I have a problem very similar where I am backing up the backup drive on my SQL server, and it causes major issues on the SQL side, with clients getting time out errors etc.

Inside my backup job I have told it to only backup the drive in question, but I think the issues come from the removal of the snapshot on the VM side, how do you prevent this from happening?
On the EDIT "hard disk" settings of the VM in question, MENTIONED ABOVE there is an independent check box and some radio buttons. One would assume when this is checked that it does not get affected by a snapshot. HOWEVER....

2 jobs: one being a full and one being only the backup drive....How are others handling this?

[Vitaliy S.]

but I think the issues come from the removal of the snapshot on the VM side, how do you prevent this from happening?

Yes, you're absolutely right. Please see the second thread you're posting to for the possible resolution.

2 jobs: one being a full and one being only the backup drive....How are others handling this?

Why not to hit your SQL Server with only one backup job? In both cases your full backup will have the second drive (the one that holds SQL backups) backed up as well.

[Wlacroix]

Yes, you're absolutely right. Please see the second thread you're posting to for the possible resolution. Why not to hit your SQL Server with only one backup job? In both cases your full backup will have the second drive (the one that holds SQL backups) backed up as well.

ITs physically too much data, the powers that be want 15 min backups on the SQL backup drive.

C drive is 30 gigs ish, a standard windows drive.
D drive is 500 gigs of transaction logs
E drive is 800+ gigs of databases

This server takes a good 14 hours to backup with compression. We are moving about 1.2 gigs of data per min raw.
There about about 300 gigs of changes per day, with backups and what not we would have to move that 300 gigs all the time. We were running a single test on a single database.
Backups are full daily, diferentals 3 times a day, transaction logs every 15 mins. We need to capture these backups to veeam some how. Its a fair chunk of data. If its running every 15 minuts its not so bad.
We are going to try removing the other drives from the snapshot feature inside vmware and start the 15 min job again.

We do a full every quarter or so for peace of mind.

[dellock6]

Uhm, maybe you can try a different approach...
Just an idea: have you tested a maintenance plan on those databases saving backups locally? If they are faster and less stressful on the database than Veeam, you can add another drive to the VM, save all the backups here, put all the other disks in indipendent so they are not snapshotted nor saved during Veeam backup, but since you save also the backup disk, you end up having the SQL backups stored in Veeam.

There are some situations where the IO is too much to allow a frequent backup rotation like 15 minutes, you hit one of the problems beeing the high daily change on data blocks, another is the snapshot commit at the end of every backup run.

Luca.