@agowa338 Lto is pretty good stuff. We delt with proprietary backup software so no help there.

For my backups, rsnapshot has worked well. But that assumes a dedicated backup server. The one I had at work was fully automated and we never lost any data that had more than 6 hrs of age over the decade or so it was running. Includes replacing the backup server and drives.

For home, it's power the server on, sanpshot, power it off, except on Fridays for software updates before powering off.

@agowa338 from a long line of failed sata and sas controllers and failed motherboards, failed drives, some running unraid - technically RAID4, I recommend only RAiD10. And not using some nas software - too much overhead for what is access to nfs and iscsi.

Also, how will you back it up? Even if, somehow, a 10 GigE backbone gets setup, 20TB is still almost 24 hrs if not more. Adding in NAS layers means taking longer.

Having spent a few decades with hardware NAS's and HA based glusterfs, keep it simple and document how to recover when it blows up and all data disappears due to some raid glitch, os fault (lvm metadata corruption), ecc errors, multiple drive or controller port fail. It will happen.

@Viss I tried to drill into my last group that Disaster Recovery mean you have a USB drive with an OS iso and brand new servers with nothing on them in a location that's no longer your lab.

And the local wiki on another drive with any security certificates and the build/release system needs to be built from scratch with just that.