adrian-golian
commented
2 years ago (perhaps it's my own misunderstanding here: ) We could consider services like loadforge, the thing is, all our backend functionality is now in managed AWS services, which provide scaling features, including auto-scaling / on-demand read-write dynamodb capacity, reserved lambda concurrency. In short, I would not mess around much with intentionally pushing the limits of out service infrastructure.
What we should be able to measure is the service performance, endpoint latency.
To deliver progress on #160 and for other purposes, we need to know how much CPU time the game engine service process spends executing different parts of our R code.