I'm starting to get curious/suspicious of whether AZs really needed - specifically with their cross-AZ bandwidth fees. It seems like when something goes down, it goes down across the entire region, rather than an individual AZ.

I know there are situations like whether you can schedule a spot instance in an AZ, etc., but for the most part it seems like the AZ is just as reliable as the region?

Curious if anyone has experienced otherwise. I'm thinking of making a site to track this for a few regions just by putting EC2 instances in each AZ and monitoring their uptime (autoscaling group for rescheduling and health checks). I know that would only cover EC2 accessibility but I think that's a good baseline monitor.