> Multiple dynos are also more likely to run on different physical infrastructure (for example, separate AWS Availability Zones), further increasing redundancy.
It's not guaranteed. As I understand it, it's random.
And as for regions. They only help you locate content to a closer region. You cannot use them for balancing/redundancy. As any data or third party add-on you use, will manage it's data in it's own specific zone. Thus creating latency issues.
I'm running a production webapp since May on the Heroku EU region, not a scratch since then.
It's a Beta suited for production apps since February:
"Beta Testers,
Heroku Europe is ready to begin running your production apps!
Create an app in the region:
$ heroku create --region eu
Then add add-ons, deploy, and scale as usual. Please note we're still adding
capacity in this area, so contact us if you expect to run more than 30 dynos or
do more than 500 reqs/second on any app in Europe.
As always, private betas should be considered confidential. Please don’t tweet,
blog, or talk about this feature publicly until we announce it ourselves.
If you need deeper control of your application platform infrastructure, you should check out something like http://deis.io/. You can run a private Heroku in any EC2 region, zones, etc.
Running your own PaaS is not for the faint of heart -- and Heroku certainly saves you time and headache -- but it's nice to have a private PaaS option that is Heroku compatible.
Sorry, what I meant was that I wasn't suggesting it was their fault, I was merely reminded of my own curiosity of wanting to build a better product but not wanting to break what a ton of customers are using and paying a lot of money for
"This post will reference the AWS services that we use behind the scenes so that we can be very specific. Note that although we will be discussing various AWS service failures, we don't blame them for what our customers experienced in any way. Heroku takes 100% of the responsibility for the downtime affecting our customers last week."
In other words, they shouldn't be exposing AWS outages to users (although as as long as they use a single cloud provider that's impossible to avoid in general.)