That's one of the great things about AWS - even though their status pages are wildly inaccurate, they are so big a quick twitter search will tell you if there are widespread problems :-)
Seriously, the big issue from yesterday's event was that even though only one AZ was affected, the official AWS APIs were constantly returning errors regardless of which AZ you were calling them from. Which means that even if you spread your estate across multiple AZs for H/A purposes, if you relied on making calls to the AWS APIs (e.g. to remap elastic IPs to fail over NAT instances for example) you were stuffed. I'm guessing that's why the big guys like Foxtel etc failed.
Luckily we rely on boring old DNS failover for our ap-southeast-2 resources - worked a treat !
Biting the hand that feeds IT
