Do you know of a good explanation for fault domains and update domains?

Question

I'm new to Azure and have been struggling with a concept specifically update domains and fault domains. Probably having a harder time understanding Update Domains. So as I understand it, having 3 VMs in 3 fault domains would be essentially having those VMs spread out to three racks? Is that correct? Like this

Fault domain 1	Fault domain 2	Fault domain 3
VM 1	VM 2	VM 3

If that is wrong, please correct me. So then what is an update domain? A lot of the documentation I have seen shows a demonstration for the fault domain similar to the table above and will describe what kind of sounds like the fault domain. If you have a link to a good explanation that would be a big help or if you think you could dumb it down for me a bit, that would work too.

Melessa

VMs don’t really exist in traditional 12U racks (but their hosts do), but you’re otherwise correct. – Sarnen

Thank you, but what do the update domains do? what are they? – Melessa

"Update domains" are when physical VM host machines are taken out of service for host OS patching, updates, etc. – Sarnen

Answer 1

Each virtual machine in your availability set has an update domain and fault domain assigned. Fault domains indicate the group of virtual machines that share common power source and network switch limiting the impact of potential physical hardware failures, network outages, or power interruptions. Update domains indicate the group of virtual machines and underlying physical hardware that can be rebooted at the same time ensuring availability of some virtual machines during a planned maintenance. Link: https://learn.microsoft.com/bs-latn-ba/azure/virtual-machines/availability-set-overview#how-do-availability-sets-work

Foreside

Thank you! Okay so in that illustration lets say all of those update and the 30 mins pass. Does the update domain just cycle to the next VM in the fault domain to apply the update? – Melessa

According to the documentation, it does cycle through the update domains of all fault domains but not sequentially. So, after the 30 mins recovery time, it will start maintenance of a different update domain from any fault domain. – Foreside

Thank you very much. Im taking a course in microsoft learn and this page was not in the materials. – Melessa

If all the VMs in Fault domain go down at the same time, then there wont be any service down because the service requests are being handled by other replicas in other fault domains. So why cant we update all the machines in a single fault domains at once. Why are we distributing this update domain across multiple fault domains? – Penstock

For example, when more than five virtual machines are configured within a single availability set with five update domains, the sixth virtual machine is placed into the same update domain as the first virtual machine, the seventh in the same update domain as the second virtual machine, and so on. The order of update domains being rebooted may not proceed sequentially during planned maintenance, but only one update domain is rebooted at a time. – Foreside

Answer 2

Under certain circumstances, two VMs in the same availability set might share a fault domain. If all the servers part of fault domain are restarted there is fair chance of your application going down if fault domain is restarted.

Elation