As we are all (I hope …) painfully aware, decent backups are absolutely
vital. Last week I looked at a potential option for offsite backup, and
solutions like Bacula are excellent for
providing a stable, reliable onsite backup, with historical data available.
Offsite backup and historical backup have their place, and their limitations. When it comes to rapid access to backed-up data in the event of a major disk crash, an instant slot-in replacement is needed, and rsync is well set for the task.
But these solutions don’t provide rapid access to the backed-up
data in the event of a major disk crash (e.g., if a centralized home
disk bites the dust, leaving everyone on the system unable to work).
What you want here is an instant slot-in replacement, and for this,
rsync is an excellent option.
rsync doesn’t keep historical data, but if all you want is a
nightly current snapshot, it does that very well. You can use a “proper”
backup solution for your historical data.
Either set rsync up onto a disk that you can physically slot in to your
regular or another server; or use a spare machine that can be plugged in
immediately instead (e.g., by changing your NFS mappings). Set up a
cron job to synchronize nightly, and you’re done. In the event
of a serious main server crash, just slot in the new disk and get everyone up
and running again with only 24 hours of data lost — which if your backup
schedule permits, you can then recover at leisure.
Be aware that although this will work if the main disk crashes overnight — as
rsync will not be able to access it so won't run — it will not work if
something more subtle goes wrong. In this case, both disks will have corrupt
data, and you're back to using your normal backups for a regular restore.