2

Allow deduplicated server backups


A
artichoke

Allow deduplicated server backups.

Currently, automatic backups take the form of *.tar.gz files. Such a file can be conveniently restored from the DirectAdmin menu.

Each backup creates a new *.tar.gz file for each user. Now if we back up our entire server to some other location, e.g., via rsync, rsnapshot, borgbase, tarsnap, or other means, each of these *.tar.gz files will be copied in its entirety.

An optimization strategy would be for DirectAdmin to save each backup not as a *.tar.gz file for each user, but as an extracted version of that file, i.e., as a directory hierarchy for each user.

This makes it a lot easier to back up the entire server. The directory containing the backup for each user would contain mostly unchanged files, so file-by-file deduplication (rsync, rsnapshot) or block-by-block deduplication (borgbase, tarsnap) would become possible, and very little data would need to be transferred to back up the entire server.

So we are balancing disk space (saved by *.tar.gz backups) with data transfer volume (saved by backups into one directory hierarchy per user). An admin or reseller should be able to pick which one they prefer.

Implementation should be very easy, since the only difference between the two options is whether or not to run the gzip + tar step during backups.

A