I would love to see a feature available when backing up VMs, where the policy would be able to look at the HOSTS that the clients are sitting on, and be able to group (or stagger) the backup jobs, such that only X number of jobs would run at a time per host.
The reason I ask is that I have backup issues on a number of VM hosts, running on an older 1GB SAN connection, or on RAID 5 local disk. Occasionally, when the VM backups run, the VMs appear to hang while the backup job finishes (certain networked services are logged as 'failing', and then return to normal when the backup is complete).
For example, I might have a VM host with 10 VM's on it, all of which are in the same backup policy. When the policy starts up, it kicks off backups of ALL 10 VMs at the same time. From what I can tell, this causes resource contention, mainly at the disk I/O level. I have limited the number of "Jobs per policy" (attributes setting), which has helped a little, but I still get intermittent service failures from those VMs during backups.
I could group the policy to spread the load across VM hosts, but we use vMotion, so those groupings would easily change over time.
Thanks
