Feature request : per-volume backup file

[inferno66]

Hello,

We're using Veeam backup (v10à with Windows to backup Filers (Virtual / Physical)

We backup to SOBR Repo (local disk + S3) with short retention policy and we backup to tape (AWS Storage gateway VTL)

Some of our files servers have huge volumes (5/10tb)

We created a backup job for each volume of each server so if needed we can retrieve from VTL only one VBK corresponding to the needed volume.

But as only one Windows Agent backup job can run in the same time on each server, the backup speed is not the most optimal.

We tried to create a backup job containing multiple volumes, the backup speed is much higher but the problem is that it's generate only one VBK file (so if we put three 10tb volume on the job, the VBK file will be 30tb which is really to big if we need to retrieve only one file from VTL)

We tried to enable "per-vm backup file" on the repo and the SOBR but it seems that there's no effect.

Is it possible to include the possibilty to make "per-volume backup file" ?

Regards

[HannesK]

Hello,
as far as I see, your goal is to have granular restore without massive restore costs / time from the VTL?

The my question is, what the reason for the VTL is? Writing data to object storage with capacity tier is much more efficient (massively less space usage = cheaper) and allows granular file / object restore.

https://helpcenter.veeam.com/docs/backu ... ml?ver=110
https://helpcenter.veeam.com/docs/backu ... ml?ver=110
post338749.html#p338749

Best regards,
Hannes

[inferno66]

Hello,

We already store to S3 capacity tier.
But it does not allow to store to S3 Glacier Deep Archive

Now with Veeam V11 you can store to archive tier (not tested yet on our side), but from that I read on the Veeam V11 doc, you still cannot make granular restore, you still need to retrieve the whole Veeam Backup file is you only need to restore one object

Regards

[HannesK]

Hello,
hmm okay, that makes sense to me... just because I'm interested: are you doing file level restores from many months old backups regularly (because for short term, some weeks, few months, Glacier is more expensive than S3 / S3 IA)?

Anyway: there are only very little amount of requests for that compared to other requests. Your request is counted +1

Best regards,
Hannes

[inferno66]

The fact is that for limiting storage cost on AWS we are only using short retention on SOBR (one week on EBS local volume / 2 weeks on S3 capacity tier), so older backup are on glacier.

We don't have many request for restoring objects older than 2 weeks but theses restore are always file level restore (so restoring 30tb just to restore a 1Mb file is not so great ^^)

Regards

[HannesK]

Yes, single file restore from tape is also a "classic request". That would also solve your specific issue because you use a VTL.

If you put everything older than two weeks on glacier, then you probably pay more than just keeping it on S3 / IA (depends on your retention).

[inferno66]

Hello,

Any news regarding this features ? :

- Per volume backup file : in order that a job create a vbk file for each volume
- Single file restore from Capacity / archive tier

REgards

[HannesK]

Hello,
- no short term plans for per-volume backups
- that's already possible today (assuming that you use VBR with capacity tier / archive tier)

Best regards,
Hannes

[inferno66]

Hello,

Exact for FLR from capacity tier.

But for archive tier we've not yet implented yet (just S3 Glacier VTL), but on the doc (https://helpcenter.veeam.com/docs/backu ... ml?ver=110) from my understanding, we need to retrieve the full archive backup file from archive tier before restore an indiviual file.
Or is it possible to restore directly only one file (for example a small .docx file) from a 8Tb archived veeam VBK file, without retrieving first the backup file?

Regards

[HannesK]

sure, the storage class has to be changed (meaning the restore point that contains the file you try to restore). I mean, the data is on tape at Amazon... it's not that a software can access it directly somehow. But you don't need to download the 8TB to your local storage.

The idea of archive tier is "write once, read (never)". If you plan to restore regularly from archive tier, I recommend to re-think the design. Infrequent access (or just a cheaper cloud provider) might solve whatever you try to achieve. cost structures in AWS (and Azure) forbid what might sound "easy"

[inferno66]

Because actually with S3 VTL, we need to retrieve the entire VBK file to local storage (and our backup job are about 8Tb each), so it take some take time, has important retrieve cost and we need to have the free space to store the backup file locally

But OK with archive tier it'll retrieve the full vbk file to capacity tier in order to restore a file.


No we don't restore regulary, (we had to do it maybe 1 time in a year from VTL), but as this a backup, you never no when you'll have to restore ^^

[inferno66]

Hello

With the new Veeam V12, is it still not possible to configure "per volume backup" ( in order that a job create a vbk file for each volume) ?

Best regards

[HannesK]

correct. there are no plans in foreseeable future for "per volume backup chains"

[jkolson]

I'd like to also put my name down on this feature request. This would very helpful to backup large VMs in general, and from my experience to help in the event of an issue with the backup chain. Veeam's answer to 90% of every support case I've opened is "take a new active full", which is painful when dealing with a large VMs when the problem resides on an individual volume. It would be nice to only really need to take the full on one volume. I too accomplish this today by just breaking out my job for each volume, and take the hit at not being able to run in parallel because the latter is much more painful.

[DataDefender]

I also strongly support this feature request. We have similarly divided our volume backups into separate jobs, as we've found that this significantly speeds up the restoration process from tape.

[dmohlmaster]

I'd like to push this one as well.
We have multiple fileservers with multiple volumes backup up using the Windows agent. One fileserver in particular has a 30TB backup file for about 7 volumes. Even on local storage, mounting this backup to retrieve a file takes nearly 5 hours. Offloading the backup to Wasabi is all or nothing, and synthetics can take over 24 hours. I'm in the process of breaking this fileserver out into separate jobs per volume. Allowing the job to save each volume into a separate file as my Hyper-V backups do would help to increase stability and ensure I can get my backups off site in a more timely manner.

[Gostev]

What makes you think this will make any difference to the time it takes to push the exact same amount of data off-site into Wasabi, or "increase stability"?