Performance tweaks for single restore point replication?

[colohost]

Hi all, I'm configuring replication (version 11) jobs and am finding the snapshot consolidation on the target end consumes the majority of the replication job execution time. I found the forum thread on how to raise the concurrent snapshot deletions per host and per cluster, which has helped, but it's still spending a lot of time on snapshot consolidation, and it doesn't begin that part of the process until all the VM's in the job have had their changed blocks updated. If it began the consolidation per-VM as they completed updating, that would leave a lot less at the end still needing to be consolidated.

I'm only keeping one restore point, as these replication jobs are part of a data center migration, so this is not an issue of too many restore points. I only need the target images to be near current to allow for rapid planned failover when each VM's time comes.

Was curious if there are any options for replication that bypass the consolidation part on the vSphere side if you only need that one restore point, or some way to do it direct to storage where only changed blocks are written into the vmdk, etc. I've been digging around and haven't found any.

[HannesK]

Hello,
from Veeam side the only replication idea that comes into my mind with V11 is using CDP. It also needs to consolidate, but the load should distribute better than with snapshot consolidation.

For a one-time migration, Veeam quick-migration might be an alternative: https://helpcenter.veeam.com/docs/backu ... ml?ver=110

Best regards,
Hannes

[colohost]

Thanks. I'll keep tinkering with it and reply if I find anything that may help with my scenario. I can't do quick migration because the initial VM size is pretty large, so replication and that final copy + consolidation is still going to be faster. CDP probably won't work well due to the level of I/O and number of VM's.

I just noticed in the docs that a planned failover creates an additional restore point, so I think I need to spin up a test VM and try a planned failover, either with or without bumping the restore points from 1 to 2 after the most recent scheduled replication. Perhaps with some combo of those, it will just write the changes to a new snapshot, not delete the oldest restore point, and allow the planned failover to boot the target without any time spent on the vSphere consolidation side, which would make the process much faster. Then I can finalize and consolidate later once the VM is already online.