Availability Zone Failure

Availability Zones are a collection of many things: networking, management components, compute, storage, etc. - and all the other parts that tie them together. So it's very difficult to say "what happens during an Availability Zone failure" because - what's the failure you want to talk about?

If we assume "the Availability Zone has been wiped off the map and is never coming back" then it's clear what the answer is. :-)

But what if it is just a temporary network connectivity failure? That doesn't affect the EC2 instance state at all because nothing has happened to it. For an external observer (outside the Availability Zone) the instance is not available but it's actually running fine.

And naturally there are many other failure modes with many other answers.

So it's much better to look at the problem from another angle: What do you need the availability of the application to be? What components make up the application? How can I design my AWS infrastructure to ensure that I meet my availability and uptime requirements? At this point, thinking about the failure of an Availability Zone is much like other failures such as "what if the network is unavailable" or "what if the instance fails". And those are scenarios that we provide answers for: auto-scaling or networking that automatically routes around problems.

But as above: It's not a question that there is a single, easy answer to.