Another Thread on Snapshots

[StorageEater]

Hello,

I've got a few questions on how Snapshots are handled with FreeNAS. I'm using an iXSystems FreeNAS Mini running (FreeNAS-9.3-STABLE) in case that is useful to know.

I've read many threads on Snapshots here and one thing that I dont see mentioned is how good of an idea it is to be running from a Snapshot. By this I mean, taking a Snapshot of a given Dataset on day 1 after loading all your data onto the box and then from that point forward using said Snapshot as your primary storage location point going forward. Days, Weeks and Months go by as the Snapshot accumulates new data while the original point in time is left alone. I ask this question as someone who deals with Storage systems and VMWare in my day job where running a system from a Snapshot is never considered a good idea for long periods of time. So why does it seem OK to do in the FreeNas world? Does it work differently?

Also, one thing I haven't seen (or been able to find) mentioned in the documentation of FreeNas is the option of "Committing" the Snapshot so that all data from the Snapshot is merged together with the original point in time when the Snap was first created as a way to get rid of the Snap while also retaining the changes made to it while running in that "Snapped" state.

Thanks for any info you can provide.

 

[anodos]

I recommend reading cyberjock's zfs primer. ZFS snapshots are not the same thing as VM snapshots in VMware.

Also, as you mentioned, taking a single snapshot after moving data to the server is not a great idea. Create snapshot tasks with a reasonable retention period.

 

[Stux]

Snapshots are read only. Clones are derived from snapshots.

It's fine to continue to write to a volume which has a previous snapshot

 

[StorageEater]

Thanks for the info all