In learning how to scale our Azure virtual machines we must first understand what Microsoft calls Fault Domains and Update Domains.

If you were to imagine an image of a Datacenter which has what we call a “rack” of servers, or a group of servers all connected to the same power and networking equipment, then this would be what is referred to in Azure as a Fault Domain (FD). So I have a server on a rack, so what? Well, imagine you are managing a web service which you provided across two different web servers that happen to be on two different racks and thus two Fault Domains. Because each rack shares the same power and network sources separate from other racks, if one of your web servers in one Fault Domain were to go down due to a power or network failure, you still have the other rack available that houses your second web server and your web service continues to operate. To guarantee that this happens you’d have to put your two web servers in what’s called an Availability Set which we will learn about shortly. Azure has an 99.95% SLA for availability of your VMs as long as they are in an Availability Set.

Now let’s contrast this with Update Domains (UD). Just like on-premises there is sometimes a need to bring a server down for reasons such as needing to patch the server or install updates on the underlying host in which case you need to purposely bring down all your servers, however, you’d need to bring them down in a staggered approach. Using Update Domains allows you to accomplish this. You should know that Microsoft does all this for you if you’re using Azure as Platform as a Service or Software as a Service, however remember that in our case we’re using Azure as Infrastructure as a Service, and thus we not only have full control of our VMs, but it is also our responsibility to configure how VMs are updated ourselves.