If you use ZFS (all of
my machines do this) you can
replicate the virtual disks of a VM to the other two nodes. Replication can automatically run once a day down to once every minute. But
not every second. So the "copy" is
nearly identical, but not 100%.
In the case that one of these disks fail then all data on
that specific disk is lost. But the other two have copies - dated from the the last replication. Now you replace physically the damaged disk (for which you have to follow some steps, just connecting power and SATA is
not enough) .
Storage Replication:
https://pve.proxmox.com/pve-docs/chapter-pvesr.html
Storage Types:
https://pve.proxmox.com/pve-docs/chapter-pvesm.html#_storage_types
PS: personally I do establish local redundancy via ZFS by establishing mirrors. When one disk dies my VMs do
not notice this ;-)