... It may surprise some HN readers, but AWS outages aren't the only reason other sites go down. Having multi-AZ just means you're resilient to a localised, single-AZ AWS failure. It doesn't help for the 1000 other ways your service could go down.
An ex-github employee who used to give talks on Github scaling challenges revealed they shifted from Rackspace.
See yourself - https://github.com/holman/ama/issues/553
Do they use AWS or another commercial cloud provider, or do they have their own servers in data centers (hopefully scattered around the globe)?
If AWS, are their services spread among multiple availability groups? I'm just wondering how this could happen.