I find it rather odd to have a GLOBAL 500 even for a minute (it may be still lasting at least in some regions http://downdetector.com/status/youtube) when a web app is distributed among servers, services, dockers or whatever and has so many systems in place to roll back automatically.

The same thing happened about a week ago with Fb Messenger (http://downdetector.com/status/facebook/news/53463-problems-...).

I expect this level of outages in 2016 to be only caused by solar flares, am I missing something?