[jaggederest]

I've been saying this for a while.

If you are spending less than $25k/mo on Heroku or equivalent, you're not ready to move off it yet.

A lot of times when I talk to people with a high Heroku bill looking to move to bare metal, I end up being able to optimize it by 1/3rd or more just by picking dyno types, scaling appropriately, and consolidating workers. You can't really do that when you're hiring headcount.

[icebraining]

You can't really do that when you're hiring headcount.

I'm curious, why not? Can't you do it on a regular basis, like refactoring?

[jaggederest]

Because "refactoring"/"consolidating workers" in headcount means firing people, and that's pretty cruel to do on a pure cost-cutting basis in a small company :)

[grx]

How can you compare the monthly price versus VM types/capabilities?

What does my bill say about the performance of my app in an e.g. Amazon environment?

[albertgoeswoof]

he/she means you can probably get away with just adding more dynos or whatever to your PaaS to get the performance you need, because the costs of moving out will be significantly higher (opportunity costs and time spent)

eventually if you're spending thousands per month on a PaaS it might be time to reconsider

[jaggederest]

A) you can't

B) Nothing in particular

It's not about the application or any technical concerns. It's purely a business decision - above $25k a month you have room to hire people and still save money, below that you really strictly do not.