An accessibility set consists of a set of discrete virtual machines that have their own names and individual properties, but propagate across error domains, which means that when you have multiple sets of virtual machines in a set, it reduces the likelihood of losing all your virtual machines in the event of a failure equipment in the host or rack.
A set of scales consists of a set of identically configured virtual machines that are also distributed across error domains (in fact, a set of scales is an implicit availability set with 5 error domains). The main difference is that the sets of scales are identical, so itβs very easy to add or remove virtual machines from the set while maintaining high availability, which, in turn, simplifies the implementation of autoscaling and performs operations on the entire set or subset of virtual machines. There are also API calls that support re-mapping and updating of virtual machines, which allows you to deploy the update while maintaining the service. They are useful for cloud architectures that require the deployment of a large number of similar virtual machines or must be flexible. A typical architecture may use a set of scales for agents or work nodes and an accessibility set for primary or control nodes. See https://azure.microsoft.com/en-us/services/virtual-machine-scale-sets/ for more details.
For the question of adding a new virtual machine to an existing AS, see Azure Resource Manager: move the virtual machine to the availability group and also https://gallery.technet.microsoft.com/Set-Azure-Resource-Manager-f7509ec4
source share