I have some concerns about this. Maybe I’m just not understanding it correctly. I have a fair number of computers that I want to backup that have 500GB drives of which maybe only 90GB is being used. If 4 computers happen to have full image backups running at the same time, I suddenly need to have 2TB+ of free space available when I would only need to have about 360GB.
In this same scenario it could actually increase fragmentation if the client drive is not very full. A 500GB mostly-contiguous image file would be allocated and then when the image was complete and the sparse holes were punched into it I could end up with a bunch of extents in the middle that are released.
How would this work when using BTRFS and raw cow images? Since every full image backup after the initial is really just an incremental?
Thanks