v9 Backup Copy and Restore Parallel Processing

[rreed]

Afternoon all. So in v8 we have parallel processing which backs up multiple VM's each w/ multiple HDD's at the same time. Nice, great. Backup copy and restore jobs do not. Will this be address in v9 please? Details?

[veremin]

With per-VM backup chain option enabled, backup copy will be creating several files (one per source VM) simultaneously. Thus, parallel processing will be achieved. Thanks.

[rreed]

Awesome, thanks Vladimir! What about the restores? Restoring a given VM's multiple HDD's in parallel?

[foggy]

Restoring a given VM's multiple HDD's in parallel?

This is on our radar.

[rreed]

Excellent, looking very forward to it! This would score a major win for Veeam and RTO.

[bradramsey360]

With per-VM backup chain option enabled, backup copy will be creating several files (one per source VM) simultaneously. Thus, parallel processing will be achieved. Thanks.

Will this also apply to Backup Copy Jobs to a Cloud Connect repository?

[veremin]

This will not be available for Cloud Connect repositories in v9, however we are planning to enable it in the next version.

[lennis40]

With per-VM backup chain option enabled, backup copy will be creating several files (one per source VM) simultaneously. Thus, parallel processing will be achieved. Thanks.

So if I'm reading this correctly, the backup job must be using a repository that has per-vm backup files enabled. That will allow the backup copy job to use parallel processing? We have done several upgrades to v9 at customer locations, and backup copy jobs aren't using parallel processing. I have enabled the per-vm on the local repository, I just haven't run the manual active full on each backup job to allow them to use the per-vm backup files. Does parallel processing only work for copy jobs running to a local repository, if cloud repos aren't supported yet?

[foggy]

Actually, direct backup copy jobs support parallel processing even without per-VM backup chains enabled, here's an excerpt from the What's New document:

Backup copy parallel processing. With v9, backup copy jobs will now process multiple VMs in
parallel, just like primary backup jobs. This improves the backup copy and retention processing
performance due to removing “dead time” between each VM, and will further speed up processing
when per-VM backup file chains are enabled on the target backup repository

Per-VM backup chains allow storages with limited ingest rate per stream (read dedupe appliances) to fully benefit from parallel processing and further improve job performance by using multiple write streams to the repository.

[lennis40]

That's what I thought, but it seems that I have some that aren't running in parallel after the upgrade to v9.

[foggy]

If they are using direct transport mode, this is unexpected. Please contact technical support for a closer look.

[rreed]

Your jobs have run a full since their upgrade, correct? And do I have that right, that they'll need to run a full backup in v9 before it begins splitting them out into separate files per VM?

[foggy]

Full is required for the per-VM chains to take effect, right. However, existing jobs should use parallel processing right after the upgrade.

[rreed]

Morning, team. V9 U1, restoring a full VM appears to be processing multiple HDD's one at a time. Has parallel processing restoration of a single VM's multiple HDD's not yet been addressed please?

[foggy]

Parallel disk restore is not yet available in the current version.

[rreed]

Crap! U2? U3??

[foggy]

Cannot share any ETA at the moment.

[rreed]

Thanks foggy. Looking very forward to it, it will really knock restores out of the park when this is available.

[marco.horstmann]

I think this question will fit into this topic. I cann't find any information how many FLR Restore sessions
can be used parallel. Is there a limit?

[foggy]

There's a limit on the number of mounted disks (though, current default value is 128, I believe).

[marco.horstmann]

Thx Alex,
VMs have only 2-3 disks. They delegate restores to the VM owners and asking how many restore operators can restore at the same time.

[rreed]

We've had three different guys running a mix of FLR and full VM restores at the same time; I myself was fully restoring two VM's while the other two guys were also restoring stuff. No issues.

[foggy]

They need to make sure the temp directory also has enough space for FLR cache. Anyway, the best way to answer is to test and check what their environment actually allows.

[rreed]

Good point, foggy. Thanks. Is the space requirement equal to the total size of what we're restoring?

[foggy]

No, it needs space for the changes performed within the mounted disk (since the disk is mounted from the read-only backup).