Cloud hosting is fantastic but it's a trade off. There are so many layers of abstraction between you and the hardware that you are completely at the mercy of one, two or (more!) technical organizations, each with their own support systems and varying levels of opacity into their infrastructure.
The fact that Amazon went down IS VERY interesting for a customer of Heroku.
And if it isn't, than that customer is a fool for outsourcing so much of their system without even understanding the risks involved.
What would be interesting to me about an Amazon outage being behind a heroku outage would be to keep a tally, and if heroku didn't manage to build in more reliability to be resilient to even an amazon outage in a particular region, to question whether they were a good fit.
In the past 30 days they have had 2 outages which have lasted more than 2 hours. That's a lot of down time.