Recovery from Tape - Needs major work

[Doug.Butler]

+1 on this for me as well.

As for whether we need "granular file restore" more or less than we need "granular VM restore" my inclination is to say "granular VM restore" because it's sort of the middle ground.

The main issue I face (as others have mentioned as well) is that I do not have enough disk space in any backup repository to restore an entire tape backup. As others have said, that is one of the main benefits of tape functionality for SMB's. I can limit the number of restore points on disk to conserve disk space, but also spin recovery points off to tape for archival purposes. Those archives are useless if I cannot free up enough disk space to restore ANYTHING from tape.

If I had to recover the whole VM backup to disk first, in order to restore a single file from the VM, that isn't too bad and again it's a 'middle ground.' As it is right now, the only use I have for backups on tape is in a DR scenario where I'm restoring EVERYTHING from tape anyway. I can't restore ANYTHING right now from tape.

[Dima P.]

Good example, thank you Douglas!

[vanakenarch]

+1, I also have limited diskspace on my backupserver, so i want to restore directly from tape and not first copy the whole tapeset to disk which takes me 13 hrs and then to restore 1 file in 2 seconds.

[TMD]

+1 - Backup to tape functionality is awesome however the inability to restore individual files/folders or even a vm without mounting the entire backup is really limiting. I understand the expectation is that any restore from tape job would likely be a DR scenario, however just to test it I need to mount a 2.4 Tb VBK.

I guess the point I'm trying to make here is that if I can't easily test/verify the tape backup periodically the functionality is essentially worthless.

Also - Regarding the Edit Media Pool wizard, not requiring the user to enter the encryption password a second time for comparison and not displaying the character (or offering the ability to display the characters) while entering the encryption password is a recipe for disaster...

[veremin]

Also - Regarding the Edit Media Pool wizard, not requiring the user to enter the encryption password a second time for comparison and not displaying the character (or offering the ability to display the characters) while entering the encryption password is a recipe for disaster...

The said control just refers to standard VB&R password manager that does allow characters display.

Thanks.

[E.V.]

+1 on this
I need to restore a 2kb file from a 80 GB VM from 6 months ago. The Backup to tape is 2,2TB and my free space is 1,5 TB.
So i can't restore this 2Kb file. It seems to be a major work to add at least a granular VM Restore.

Thanks

[lobo519]

Sooo - are we getting anything on on this in V9??

[mike.r]

As Veeam representants told us, V9 is scheduled for Q4 2015.
In official Veeam documents nothing is seen to this case.
So, what do you think are we geting on this in V9 ?
Just a guess from me: nothing!

[Dima P.]

Sooo - are we getting anything on on this in V9??

Unfortunately this functionality won’t be provided in the upcoming v9, sorry. All the major tape features have already been announced via v9 page, kindly, check it for more details regarding tapes. Thanks.

[lobo519]

Unfortunately this functionality won’t be provided in the upcoming v9, sorry. All the major tape features have already been announced via v9 page, kindly, check it for more details regarding tapes. Thanks.

Are there any plans to address this issue? So if we are luck we might get it in another year?

[veremin]

It's certainly on our radar, but in the upcoming version priority has been given to the more demanded functionality. Thanks.

[dellock6]

Also, even if it's not the solution you are asking, the new v9 will have per-vm backup chains, and this will help a lot. I'm a little sad it almost passed under the radar (http://www.veeam.com/blog/backup-storag ... te-v9.html) but this feature is a real saviour not just for storage appliances able to leverage multithreaded writes, but it will also be beneficial for restores from tapes. In your example, if you store per-vm chains into tapes instead of one large file per job, the restore to the disk will be just about the size of the single vm/chain, so you will not need to have the free space of the entire job available.

[mike.r]

This sounds not to bad. Do I understand it right, this works with any Veeam repository, not only with the named ones ?
And this will be an option set for the repository, not for the backup jobs ?

[dellock6]

Correct, it will be an option to be enabled (even on existing ones, simply the old chain will stay intact and not splitted) on the repository, transparent to the job.

[SteveV]

Hello,

I have been investigating topic of possibility of Guest OS file-based restore directly from tapes and my findings are following:
- if we do backup to disk, it is possible to restore files to inside guest OS from the Veeam backup files on disk directly
- if we do backup to disk and then we off-load this backup to tapes, the restore from tape needs to be 2-step always = first restore the Veeam backup file to disk, then restore backup data from it to the end location
- in case we would like to restore the guest OS from tape directly, the only way to do that is File to Tape backup, but this one cannot backup the whole VM, but specific files/folders/Shares only

We are running Veeam B&R 8.0.0.2030 and we are backing up 3x VMware ESXi 5.5.0 2403361 vHosts with shared HP 3PAR Storage (connected to physical backup server over FC).

Is the information above correct = is there really no way of doing VM backup to tape with possibility of Guest OS files restoration from tapes directly?
Is this functionality planned for future releases of Veeam B&R? We would really appreciate it...

Thank you for your reply!

[Dima P.]

Hello Stepan,
All information is correct. We do have plans for implementing such features in next versions; meanwhile I’ve merged your post to the existing discussions.

[aaronk]

Dev Team,

It would be extremely useful if I could restore just a single file from tape like I can with on disk backups. Having to restore an entire VBK to get at a single file seems a little in inefficient. Are there any plans to include this in a future version?

[foggy]

Thanks, Aaron. As you can see above, this was already requested and is in our plans. Currently you can consider the new per-VM backups chains capabilities coming in v9 that could help in this scenario.

[JSi]

Hello,

sorry for this newbie question, but I started with veeam tape jobs only recently.
When I try to restore VM from Tape it is 2 phase process: 1. restore vbk to datastore, 2. restore VM . Problem with this approach is that RTO for such VM is too long (depends on size of vbk and speed of storage).

Last time it takes 8 hours (cca 0.8TB vbk on slow (30MBps) storage.

It seems to me that only workaround is to make more smaller jobs but there is more related consequences - more jobs = worse dedup ratio; much more needed tape sets. Am I wrong?

Will be this architectural drawback solved in v.9 some other way? Is there some better solution than mentioned before?

Thanks for help and best regards,
Jan

[foggy]

Jan, there're some improvements in this area coming in v9, however the whole idea stays the same, you need to restore the files from tape to disk first.

[bcann]

Add me to this list. Also add:

Backup direct to tape.
Restore direct from tape.
Proper Easy GFS (Grandfather-Father-Son) Tape setup

[foggy]

The latter will be addressed in v9 with the help of dedicated GFS media pools.

[JSi]

Thanx for reply Foggy, I didnt find this thread before. I am looking forward to new v9. repository ability to granulate jobs per vm and gfs too.
Thanx for good work in Veeam.

[ijustinhk]

+1 for single file restore from tape.

[rreed]

+1 for restore direct from tape please. I realize granular file from tape might be a large request, but at least let us go directly from tape to restored VM please. Otherwise our RTO remains double+ having to go from tape to .vbk, then from .vbk to restored VM.

[veremin]

Starting from version 9, there will be per-VM chain option. As the result, you will be also able to restore particular VM, when restoration need comes. Thanks.

[rreed]

Directly to a working VM w/o having to go to a local /vbk file, THEN restore that to a working VM?

[veremin]

I must have hurried up, while reading your last reply.

Nope, it would be still two-step operation (to repository -> to production). But we have this request (direct restore) tracked.

Thanks.

[rreed]

Awesome. Many thanks guys!!

[riconeo]

+1 direct restore from tape request too.