I'm trying to set up two VMs that are load balanced and need physical separation to ensure that if one host server or rack goes down, the other VM remains unaffected. I have some questions about configuring the availability set. I get the concept of fault domains, but I'm confused about why there's an option to select different numbers for it. Setting it to 1 seems counterproductive because it offers no resiliency, while 2 or 3 makes sense. Why are those options even available?
On top of that, I don't understand the point of having multiple update domains. With just two VMs in an availability set, wouldn't it be logical to update them one at a time? Why is there a range of options from 1 to 20 for update domains?
I only need these two VMs for a small web farm right now, but I want to make sure I can expand in the future if needed. One last thing: once an availability set is set up, I can't change these parameters, so it's crucial that I get it right. Any advice would be greatly appreciated!
2 Answers
The reason they allow customization for fault domains is to give you flexibility based on your specific needs. Choosing 1 could be for tests or very limited scenarios, but generally, more fault domains offer better resilience. As for update domains, having multiple allows you to control how updates are rolled out—useful if you expand your setup in the future. You might not need it now, but it could come in handy later!
Have you considered using availability zones instead? They offer better resilience since they’re designed to be isolated from each other, which might fulfill your needs better than just an availability set.
Availability zones aren't available everywhere, unfortunately. It's something to keep in mind.
Yeah, I thought about that, but we rely on a bunch of appliances like Veeam and a single network for our VPN. Moving to a different region would complicate things with networking, so I’m thinking sticking to availability sets is better for now.