+1 for PBS and it’s dedup capabilities. I run a remote sync with it to offsite, along with ZFS reps of the underlying datastores.

As for Proxmox itself, I haven’t bothered with backing the nodes themselves up, it’s so simple to set up and cluster that if it went down, it would be a good chance for a nuke and pave, and restore VMs.

That person went a little overboard imho. Would be interested if they ever tried a complete recovery on that setup.

Ceph hasn’t been mentioned in the article.