When a virtual machine is deployed in Azure it can be configured to be either part of anavailability set or to have the availability set configured to ‘None’. The ‘None’ option can be slightly misleading because everyone would like their VM to be highly available, the VM you are deploying is not part of a set of VMs (i.e., 2 or more) that provide the same functionality, such as multiple web servers under a common load balancer or two SQL Servers that are replicating data.
when VMs are deployed to an availability set the Azure platform ensures that only a subset of the VMs in the availability set are taken down for platform maintenance at any one time. So when you have multiple VMs in the availability set you will always have at least a portion of those VMs running. When a single VM instance is not part of an availability set you are informing the Azure platform that this machine is not part of a high availability set, and therefore we have special processes in place to inform you via email before the VM is rebooted for platform maintenance operations.
To read more about this click here,