Why DRBD when something like `sheepdog/ceph` is available?
The latter offer much better scalability and features.
MCVirt was started with the view of a small (2 node clusters), with the aim of a simple setup... We have investigated other storage backends, but felt that DRBD provided the features that we required and also, in some instances, is easier to recover in disaster recovery situations.
That said, I can't see why it wouldn't be too hard to implement ceph support in the future... we currently support both DRBD and local storage, adding a third backend wouldn't be too complex. I had also been considering other shared storage backends, such as iSCSI and, possibly, NFS