If my understanding is correct, the use of Availability Set would be beneficial if the Virtual Machines (VMs) run the same workload. Furthermore, if I am not mistaken, the scope of an Availability Set is limited to a single physical Data Center rather than multiple Availability Zones (which span multiple Data Centers). Therefore, in which cases would we require a Proximity Placement Group if we have the Availability Set?
Proximity Placement Groups enable users to keep their virtual machines (VMs) in the same data center in order to optimize performance. These groups are allowed to contain Availability Sets.
Creating a single Application Service (AS) will automatically place it in a single datacenter. However, when creating a second AS or Virtual Machine (VM), there is no available option to select the same datacenter as the first AS. Placement Groups (PPGs) can be used to guarantee the second AS is in the same datacenter as the first.
High-speed connections between nodes in clusters can provide a boost to performance for specific workloads, such as big data processing through Hadoop clusters and genomic sequencing. This allows for fast data exchange, resulting in faster processing with minimal latency. It is possible to take advantage of accelerated networking (available on supported OS and VM sizes) within the proximity placement group to maximize communication efficiency between nodes.