I'm trying to work out if previous datacenter failures on AWS have just applied to a single "availability zone" or the whole region. Does anyone know the details of the extent of the worst AWS failure?
Amazon AWS has suffered at least two major outages in its life.
The first, in April of 2011, affected one of the availability zones in US-East.
The second, in June of 2012, affected one of the availability zones in US-East. During this event, for a short period of time it was not possible to start new instances in any US-East availability zone.
Both failures were pretty bad, though they seem to have resolved them pretty well overall.