restore multi-stream from storeonce catalyst.

[norawee]

Hi all,

First of all, i want to improve performance of restore data from storeonce catalyst and need confirmation or suggestion that my solution is ok.
In my case. we have one VM that consist 3x vdisks (total capacity is around 2-3TB) and veeam repository is Storeonce catalyst's store.

In backup process. It's quite very fast because veeam can backup vdisks in parallel as multi-stream.
but when we're restore this vm, in veeam 9.5 gui show that veeam can restore vdisks in parallel as multi-stream but in Storeonce catalyst gui show only one stream and limit restore throughput. Therefore it's slower than we're expect.

My question is -- Can i split each vdisks to multiple job and restore each job in parallel and then mount vdisks from multiple job to vm when restore complete ?
If yes, In storeonce catalyst gui will show multi-stream as multiple job and improve restore throughput performances.

need confirmation or suggestion from veeam specialist that this solution is ok.


Thank you in advance !!

Cheers,
Norawee S.

[foggy]

Hi Norawee, indeed, disks are written to datastore in parallel, however, read from StoreOnce Catalyst repository through a single connection (StoreOnce does not support shared read). However, I'd warn you against splitting the disks into several jobs - this is not recommended from FLR perspective. Plus, consider the fact that VM disks will be backed up with some time shift, not quiesced within a single job. Also, I'm not even sure multiple concurrent restores to the same VM are possible.

[norawee]

Thanks so much.. foggy.

Actually, in backup process -- we use one of "backup job" for create first set of backup data and then create one of "backup copy job" to create 2nd set of backup data. therefore it should not have any issue about the time shift.

in restore process -- we will restore full vm (exclude vdisk) from serveral job in parallel. there for in storeonce will show multi-stream from 2 job. and then manual reconfig disk in vm again after restore vm complete.

pls help suggest again....thanks so much.

[foggy]

By time shift I mean the fact that by backing up VM disks in separate jobs you're getting their state at different points in time (no way to sync jobs so that they take snapshots simultaneously for all the disks).

[norawee]

Understand your point.

But i mean that i'll create one backup job for backup full vm (not split vdisk to several backup job) and then create backup copy job for 2nd set of backup data.
I will split vdisk in restore process by parallel restore full vm (exclude vdisk) from serveral job. Then manual reconfig disk in vm again after restore complete.

thank so much.

[norawee]

hi foggy

Could you pls advise again.
if i use one backup job (full vm) and one backup copy job (for make 2nd of this backup set)
and restore vm by split vdisk01 from backup job and another vdisk from backup copy job in parallel. Then reconfig disk to vm after restore complete.
can i use this way for improve restore time from catalyst store or any suggestion.

Pls help advise.

[veremin]

can i use this way for improve restore time from catalyst store or any suggestion.

I don't think you can, as parallel restore of multiple VM disks spanned across several jobs is not possible. Thanks.