A sound backup strategy has a dedicated shared storage for the backup images instead of local storage or storage that is used for the disk images themselves. This way, we can centralize the backup location and restore them even in the event of a Proxmox node failure. If the backup is stored locally on the same node, during hardware failures, that node may become completely inaccessible, causing a VM restoration delay.
One of the most popular options for a backup storage node is NFS. In an enterprise or mission-critical environment, a cluster with built-in redundancy dedicated to backups is a recommended practice. In smaller environments, good redundancy can still be achieved using storage options, such as Gluster or DRBD. With the addition of ZFS and Gluster in Proxmox VE, it is now a viable option to turn a Proxmox node into a backup using ZFS and still manage the node through the Proxmox GUI. Unfortunately, we cannot store backup files on the Ceph RBD storage...