We find currently implemented file-level backup strategy as very inefficient. We have a lot of customers with big disks (>200GB) full of small files (mailboxes for example) and tar'ing such disks takes 12+ hours to complete. We tested dd of the same snapshot backup LV and it lasts almost 10 times less. File system consistency and health check (which is inherited in file-level backup) can be achieved as well in block-level (check fs before dd, etc).
Please sign in to leave a comment.