Best practice for replicating a Time Machine share

[keylevel]

My system is used in a home/office environment, mainly as a backup target for PCs and Macs. Backups either use Acronis (full / incremental) or Time Machine.

The pool containing the backups is then replicated on to a set of external drives that are used for off-site backup, with the backup disk being rotated off-site once a week. The replication uses a "Replication Task" to duplicate snapshots of the pool that are taken hourly, daily, weekly, monthly.

However the Time Machines backups include a similar snapshot strategy produced by Time Machine itself.

Does this mean that I only need to store a smaller set of ZFS snapshots for a Time Machine share (e.g. two, to recover if Time Machine creates an invalid backup)?

 

[dlavigne]

Did you determine the answer to this?

 

[keylevel]

Not yet, but I've not really had much time to think about it or investigate further.

 

[Glorious1]

I don't think you can get into the innards of Time Machine to make your snapshots more efficient.

Personally, my philosophy is, the Time Machine dataset is already a backup. We always backup our macs also with a local external disk, either with Time Machine or a SuperDuper clone. I don't think I need a backup of a Time Machine backup when I already have an alternate backup.

That said, there is maybe a reason for it. If you ever have Time Machine tell you the backup is corrupted and it needs to wipe and start over, you would have the possibility of going back to a snapshot of the Time Machine before it was corrupted. I don't think I've ever done this, but . . .