I currently run an OwnCloud server with about 25 accounts, 2.6 TB, and moderately growing. As data will be stored for the next several decades, the OwnCloud data is on a mirrored ZFS file system, to preserve data integrity. I use rsnapshot to retain nightly, weekly, and monthly snapshots on an 8 TB drive (ext filesystem), which is periodically swapped with another 8 TB drive kept off-site.
The simplicity of attaching the 8 TB drive to any linux box is appealing for file or system recovery. This has been working well for 15 months. Have not yet needed to restore from backup, but 2 failing drive were swapped out on the ZFS.
Is there a significant advantage in using ZFS snapshots and/or using ZFS on the backup drives for improved file integrity? What would be “best practice” or should my current system suffice for now and the future?
Best Answer
ZFS
send/recv
is "change-aware": this means that only changed block are transferred on subsequent backups. Compared to something asrsnapshot
, which needs to walk all metadata to discover any potentially changed file, and then needs to read all modified files to extract any changes,send/recv
clearly is way faster. Rather than reinventing the wheel, I suggest you to give a look atsyncoid
to schedule regular, incremental backups.That said,
rsnapshot
is a wonderful piece of software which I extensively use whensend/recv
is not applicable (ie: when destination runs on something different than ZFS and/or I need to attach it to non-ZFS capable systems).