Replication Job - slow helper snapshot removal

[mongie]

Hi All,

We've been using Veeam for backups for years, but have just started using it for replication. I'm finding that when some of my replication jobs run, the helper snapshots can take a very long time to be removed.

In this case, 503GB was replicated on a 3TB VM (windows file server with dedupe enabled - a long story), and the helper snapshot has been removing for nearly 15 hours. I understand that this is largely down to underlying storage performance, but I'm wondering if that sounds normal? We're also using ESXI 5.5 with vSphere 6.5... I'm wondering if snapshot removal performance has changed much in ESXi 6.5?

Any suggestions would be appreciated.

Thanks

[Vitaliy S.]

Hi Alex,

Yes, vSphere 6.x should provide better performance for the snapshot consolidation process. Please check out this post for more info.

BTW, what are the datastore performance rates when VM snapshot starts the commit operation?

Thanks!

[mongie]

What do you mean by "datastore performance rates" ?

[nielsengelen]

IOPS usage, latency (read/write) and more important how is the storage underneath (is this a SAN or a lot of disks in a certain RAID set)?

[veeeammeupscotty]

Hi All,

We've been using Veeam for backups for years, but have just started using it for replication. I'm finding that when some of my replication jobs run, the helper snapshots can take a very long time to be removed.

In this case, 503GB was replicated on a 3TB VM (windows file server with dedupe enabled - a long story), and the helper snapshot has been removing for nearly 15 hours. I understand that this is largely down to underlying storage performance, but I'm wondering if that sounds normal? We're also using ESXI 5.5 with vSphere 6.5... I'm wondering if snapshot removal performance has changed much in ESXi 6.5?

Any suggestions would be appreciated.

Thanks

I'm noticing similar behavior on ESXi 6.0 U2. About 120 GB of replicated data and snapshot has been removing for 4+ hours so far. The SAN is dedicated for backup and it's practically sleeping from an I/O and Transfer Rate perspective (~30MB/s read/write while snapshot is removing vs the typical 300-600 MB/s we get during normal backups where the proxy is bottleneck.)