I have a 2012 R2 server with a 2012 R2 guest VM. If I do a backup of the VM from the host level and a backup of the VM from within the VM there seem to be some differences.
The host level backup seems to always be a full backup - even if I have set to Fast in performance settings - whereas the backup of the same VM from within the same VM on the same day shifts from full to 'fast'.
As well there is a difference in files transferred/backed up reported. So let's say 10 gig has been modified in the VM by users. The internal backup will accurately reflect that; however on the same day the host level will show a much greater number of data transferred.
As the host level backup treats the VM as a single file VHD - or so I believe - does that account for the discrepancy? In other words even though only 10 gig of data has changed, somehow as far as the VHD is concerned it's a larger amount because of the way that VHD's account of their space. in a similar vein, could it be that the blocks changed on the VHD - which is what the backup I am assuming is looking at - is quite different than the amount of data actually changed - in other words if we add 10 gig of user data, the VHD will be shuffling a lot of blocks around to accommodate that/link it up so that more blocks are affected than 10 gig? Also would it make a difference if the VHD was dynamic versus fixed, VHD versus VHDX for any of this?