What performance/capacity metrics are you considerned about when deploying a phoenix application and how do you determine what your deployment is capable of handling?
Also, any recommendations on how you guys are deploying your applications?
Any input is appreciated.
I find the only reliable way to gauge capacity is by deploying the applicatin and then performing load tests to see how it works in practice. It’s very easy to get estimates wrong so real data is better.
I’m sad to admit that’s about the best I’ve come up with as well.
In Ruby, I could pretty accurately tell my server need based on the number of “slots” available, the average time per request, and the throughput.
I suppose you could do the same in Elixir, because there is a maximum throughput on cowboy by default, but the parallel and preemptive nature adds a lot more difficulty in this calculation.