We cannot even restart/turn off the dynos or get into the dashboard to turn off and kill our background tasks for some of our clients.
Also Heroku: we auto restart your dyno every day!
So now my app is down because Heroku forced it to restart, and none of our hourly employees can work :|
As a PaaS I would think that they would run a high availability cluster on at least 2 multiple regions so that they would have a mechanism in place for events like these. I know it's expensive, but if you charge 250 for 2.5GB of RAM I believe you would have enough money to cover it. I also think as you hinted that they should separate services across different regions..