Backup Copy to Data Domain, per disk delay

[ferrus]

We have an 18.5TB Exchange VM with 27 attached disks.
The Backup to local disks works fine, but the Backup Copy to a DD2500 is painfully slow.

Part of this is due to the size of the VM, and the fact that Veeam/DD doesn't backup disks within a single VM in parallel.
But the thing that adds most to the length of the job, is a 33-36 minute wait at the start of every new processed disk - before the actual data starts to be backed up.
That adds up to 14 hours in the BCJ without a single KB being backed up.

I can't remember whether this was always the case, or if it was due to a config change (we originally misconfigured the job to merge - which itself led to job durations >24 hours).
This morning the delay appears to be over an hour per disk, and it's only transferred 13GB after 15 hours.

Our other Backup Copy Jobs don't appear to have the same issue.
Is there an obvious misconfiguration here?

[ferrus]

Should add that this affects tiny 5GB disks, the same as 2TB ones. No difference in the pre-processing time between them.
This morning for example it was over an hour to feedback the size of a 5GB drive, before the backup started - and there was no data to back up.

[nefes]

Have you opened support case on this? It is hard to understand, what is taking time, without digging the logs.
30-60 minutes before each disk is certainly not looking expected.

[ferrus]

I'll open one now.
Silly to think I didn't notice it only happening to this one VM, but it's so massive in terms of disks numbers and space, that it's the only one still running into the next day.

[ferrus]

Always like to post a comment on threads I start, after Veeam support resolve an issue.
The problem mentioned above, which has caused huge disruption to us, was eventually solved with a very quick and simple registry key.

I won't post it here in case it's not suitable for everyone (I've since found a few references to it on this forum with different values), but suffice to say the time it takes to process each of the 20 disks before data is backed up - has fallen from a max of 50 mins per disk, to just a few seconds, shaving up to 20 hours off the backup time and allowing the job to complete.

I do wonder if the fix would be better applied per job, rather than per server - for some peoples deployments, but for us the change appears to be all positive.
Kudos to Veeam support again, who remain the best support team I deal with.

The call reference is #02897344 for anyone with the same issue.