Why Amazon is not to blame for your AWS downtime
lockergnome.com
lockergnome.com
Yes, that's what everyone was told. What the author doesn't realize is that several availability zones went down. I'm not sure the author is actually that familiar with the outage given this non-trivial mistake.
But, even if you manually pulled the record for a downed region, you can't rely on downstream DNS to respect your TTLs.
I suppose you'd at least be able to say something like "some of our users" as opposed to "we're completely down." That seems like cold comfort, though.
If you read Quora's account [1], it's clear that this was regionwide, not just limited to a single availability zone. This debunks the misguided understanding that simply having slave instances in different availability zones will be enough for ensuring availability.
While that's what you would have expected from Amazon (and the name "Availability Zones"), that's not what happened.
The issue wasn't necessarily that Amazon had downtimes--you'd be hard-pressed to find a provider that has never been down. The issue was the scope and duration of the downtime that demonstrated that availability zones mean little in terms of service redundancy.
[1] http://www.quora.com/Quora-Outage-April-21-22-2011/What-caus...