You are right that snapshotting is not supported in this case. While RDM virtual disks in physical mode, Independent disks, and disks connected via in-guest iSCSI initiator are also not supported, these are skipped from processing automatically, With VMs with disks engaged in SCSI bus sharing in this case we can't even snapshot the VM.
I am not that familiar with this but somebody else might be... Is there a way to support SQL failover cluster without using that SCSI bus sharing?
If not... I personally would use Veeam Endpoint (soon Veeam Agent for Windows) and backup the VM entirely and straight to a B&R repository so you can afterwards have all the goodies of Veeam B&R
A few things you can keep in mind:
* You can do a full backup of the VM, including log truncation of SQL by VEB, giving you (in combination with the above B&R repository) Veeam SQL explorer functionality
* You can do the SQL backup yourself and then take a backup of the entire C-volume and on the DATA volume only the .bak files from the SQL backup.
* BMR is perfectly possible, but create a new recovery media every time there are major changes to the OS (like VM upgrade or OS upgrade or similar
Hope it helps