The difference between rsync and borgbackup is exactly this: the latter does backup retention properly.
The difference between rsync and borgbackup is exactly this: the latter does backup retention properly.
This is, broadly, correct.
However, we have been encouraging our customers to configure "dumb" 1:1 mirror rsync jobs to us and then allow the platform to handle all of the retention with the ZFS snapshots.
An rsync.net customer configures any arbitrary schedule of days/weeks/months/quarters/years and then we (rsync.net) handle the rotation, retention and maintenance of these snapshots.
The benefits of this, beyond the fire and forget aspect, are twofold:
First, they are more efficient than typical retention schemes like 'rsnapshot' or "rsync snapshots" since ZFS diffs on a bit basis and not a full file basis - so if you change a file, but only by a little bit, you don't pay the entire size of the files as a penalty in your snapshot size.
Second, the ZFS snapshots are immutable. Even if someone steals your rsync.net credentials and wipes out your entire account, the snapshots cannot be changed.[1]
Of course you may do whatever you'd like - you can even configure snapshots on an account you use 'borg' with and get very inefficient "snapshots of snapshots". Whatever makes you happy - we just give you an empty UNIX filesystem to do whatever you like with.
My use case has always been to save my data in case of disaster, I haven't needed to rollback an accidental deletion... yet.
That said, rsync.net has daily ZFS snapshots with weekly retention, so you're able to roll back in that case.
This made me shudder. I used to replicate external drive A to B, in case one failed. Drive A failed by slowly corrupting file contents, which I didn't notice until a significant part of them was synced to B. Fun times.
Sync is not a synonym of backup.
For accidental deletions, 7 days is indeed most likely fine.