We have a few QCOW2 images that have grown rather large (> 250GB) and to back them up across the VLAN is resource intensive and time consuming.

What I'd like to do to optimise backups is freeze the backing store and always run the VMs in 'snapshot 1', and then run 'snapshot 2' to backup 'snapshot 1' to reciprocal failover hypervisors in the DC.

I do know it's common practice to even run multiple VMs off a single "golden image" backing store, but I'm concerned about the performance impact with such a large backing store, and 'snapshot 1' growing at ~10GB a year. With that in mind, it should be feasible to pivot 'snapshot 1' into the backing image periodically.

Further, I am aware of incremental backups which Debian 11 doesn't yet support. I think it's coming in Debian 12, but some of our servers are actually still on Debian 10.

jeff@abacus:/var/lib/libvirt/images$ virsh backup-begin dev
error: Operation not supported: incremental backup is not supported yet

In a nutshell, is it feasible to run off an > 250GB and growing backing store?

Thanks in advance for any advice.

-- 
Jeff Brown
Future Foundation
Cell 074 101 5170
VoIP 087 550 1210
Fax: 086 532 3508