SureBackup causing SCSI aborts and latency on ESXi 6 host

[bdoe]

I've had SureBackup disabled for a while, partly from moving things around and partly because the virtual lab was on an ESXi 5.5 host when I had some 6 hosts. I've since moved the virtual lab to two different ESXi 6 hosts and re-enabled the job. It runs (fails, but only on two machines) but I've found that on both hosts, Veeam ONE starts triggering a lot of alerts during the job. The alerts are datastore write latency, VM total disk latency, or host disk SCSI aborts. I never had that happen when the lab was on a 5.5 host. The application group is two DC's and a DHCP server, and the job was set to only test two machines at once - I've reduced that to one for tonight. I know the SureBackup job is using host resources, but should it be hitting the datastore so hard?

[Gostev]

Assuming you're talking about vPower NFS datastore, remember that SureBackup runs VMs from backup files, so there's on-the-fly data conversion engine that adds latency. Other datastore (the one that hosts redo log files for those VMs) should be getting minimal I/O though.

[bdoe]

Sorry, I should have mentioned that Veeam itself is on a physical host; the backup files are on a separate (also physical) Linux repository. The ESXi host has local storage, but is just a place for the lab to run. I didn't think SureBackup would be using its local datastore.

[Gostev]

Check if you have the virtual lab configured to store redo logs on that local datastore.

[bdoe]

It is. Since this is all local storage, I didn't think I had any other choice.