V9 Backup copy merge slow

[namiko78]

I notice that when my backup copy job does it's merge operation, it leaves the disk idle for most of the time. There are spikes where it will use 95% of the disk, but alot of it seems to be writing or VBM files, some kind of temp files, not sure what it's doing here, but wondering if i can make any changes to speed it up. Would per VM backup mode on the repo make any difference?

It seems like we had this issue in v8 until there was a patch that increased memory usage for metadata operations and resulted in faster speeds.

Thoughts?

[namiko78]

Forgot to include, it's a little over 100 gigs of VIB it's merging, and takes 10-12 hrs. There are some other tasks running on the server, like a nightly dedup (not targeting this job) and defrag, but disabling these has no apparent effect on disk usage. It simply doesn't work the disks very hard.

[haslund]

Is this a Windows based Backup Repository?
What kind of storage are we looking at? Disks? RAID?
How many VM's in the job?

If you have enough free space to handle it, I would recommend for performance reasons to switch to Per-VM Backup Chains on the Backup Repository.
Just beware, when you make this change you will need to run an Active Full on the Backup Job to make the switch and obviously this will take up additional space.

[namiko78]

Yes, windows.
34 disks 4 TB each, raid 6
117 vm's

Ok, I can try switching to per-vm and see if that helps.

[namiko78]

I won't know if this helped for 60 days (60 retention points) but if the disk I/O is not heavy, it should increase it's usage in per/vm mode?