If we have a single NFS Datastore and say 10 VM's in that datastore, is there any performance/gotcha's/issues with scheduling a backup by right clicking a specific VM, then choosing Netapp, then backup, then schedule a backup?
The reason would be say i only want to backup 5 of the machines nightly, instead of all 10. Any issues with doing this and not doing the entire datastore?
From the FAS appliance the snapshot is always at the volume level, so the same overhead there, although you will be creating multiple snapshots instead of one which would be minimal. The VMware snapshot overhead will be there though (vm snap, netapp snap, delete vm snap) unless you are using crash consistent which is what the default is now in VSC. Most backups of VMs we see now are crash consistent without VM snapshots (if a database then SnapManager would be a separate process).
The size will be the snap deltas for each snapshot. The oldest snapshots take the most space, so it depends on the changes in between snaps but likely will be minimal size savings if you have 10 snaps for 10 VMs compared to one snap for the datastore (assuming the oldest snap on a per vm or per datastore is around the same time). You can use the "snap delta" command to look at snapshots to compare using both methods to see how much but I would be surprised if a huge difference in size.