What happens inside Amazon when one of its AWS hosting services goes down

This is the reason that I disagree with those who say they like to control their own hardware. AWS immediately has 10s, 100s of people attacking the problem with full vendor support. Meanwhile, I'm trying to find out why my SAN firmware update borked 3 disks and HP will be on-site is about 4 hours.

One thing that AWS could do better is actually reporting these outages to their customers. It's a running joke that every time AWS does something funny, their status page stays green. Even when the status updates the indicator is a tiny light grey exclamation point; basically invisible.