Q: I understand that standalone Azure VMs not in an Availability Set have no SLA; does that mean if there's a problem, the VM is lost forever?
A: To receive the Azure SLA, it's important to deploy multiple instances of the VM and place them in an Availability Set. This ensures that the VMs are split over multiple upgrade and fault domains, which means that in the event of planned or unplanned failures, not all instances of the service are taken down at the same time—which allows Microsoft to offer the SLA for the service. If you have single-instance VMs, then the SLA doesn't apply and essentially there is no SLA. However, this doesn't mean that you lose your VM in the event of a problem. The storage containing the VM is highly available and protected. What isn't guaranteed is how quickly the healing takes place to recreate the VM on another node; however, it will come back at some point, and the only loss would be data on the temporary drive (which is always at risk on any VM and should never contain any data that you need to keep).