[DoubleCluster]

> We created open-source Puppet modules that provide compatibility with Heroku’s cedar stack (process management, dependency management, concurrency)

If I had a Heroku app but wanted to move to EC2 that'd be great. The rest of the stuff doesn't seem that interesting.

[opdemand]

I'm curious. If you have a dozen or so Heroku apps and a few peers you collaborate with, what's your management strategy?

[DoubleCluster]

Would anyone actually do that? I run stuff on AWS using their excellent Java SDK and I'm quite happy about the flexibility that brings me.

[opdemand]

Firing off API calls with an SDK is great when you're a one-man operation.

At a certain point you need deployment automation that respects infrastructure dependencies, change tracking with an audit trail, an at-a-glance view of environments... all things AWS is not very good at. Though they're trying with OpsWorks.

How do you work around that stuff?

[DoubleCluster]

Home-grown dashboard. Works fine for now. It was quite a lot of work to build it but the code turned out nice and concise and I learned a lot in the process.