So in order to explain this we need to dig a little deep in terms of how Windows Azure Data Centers are created. Basically Windows Azure Data Centers are built using “Containers” that inside are full of clusters and racks. Each of those Containers have specific services, like for example, Compute and Storage, SQL Azure, Service Bus, Access Control Service, and so on. Those containers are spread across the data center and each time we subscribe/deploy a service the Fabric Controller (which chooses based on our solution configuration where the services should be deployed) can place our services spread across the data center.
Now one thing that can happen is we need to be very careful in where we create the several services, because if we place the Hosted Service in North Central US and then the Storage Account in South Central US, this won’t be very good both in terms of Latency or Costs, since we’ll get charged whenever we get out of the Data Center. But even if we choose the same Data Center, nothing tells us that the services will be close together, since one can be placed in one end of the Data Center and the other in the other end, and so this will remote the costs and make the latency better, but it would be great to go a little further like placing them in the same Container, or even in the same Cluster. The answer for this is Affinity Groups.
Basically Affinity Groups is a way to tell the Fabric Controller that those two elements, Compute and Storage, should always be together and close to one another, and what this does is when the Fabric Controller is searching for the best suited Container to deploy those services will be looking for one where it can deploy both in the same Cluster, making them as close as possible, and reducing the latency, and increasing the performance.
So in summary, Affinity Groups provide us:
Based on this, don’t forget to use Affinity Groups right from the start, since it’s not possible after having deployed both the Compute or Storage to change them into an Affinity Group.
To finalize, and since now you can be thinking that this would be very interesting for other services also, no other services are able to take advantage of this Affinity, since neither of them share the same Container.
Nice to hear that storage stamps can share the same container than compute nodes, great explanation about affinity groups usage!
Good and concise summary, thanks.
Slight correction: the fabric controller manages the nodes of a cluster. A higher-level process controls the choice of data center and cluster. This process is called RDFE which is short for Red Dog Front End (Red Dog was the Azure project's code name). The Azure developer portal web site and web service talks directly to RDFE. Mark Russinovich has discussed this in the past. I know about it because I worked on RDFE and the developer portal before I left Microsoft a year ago.