We should have a series of easy to run load tests, maybe used as an acceptance test before releasing (eventually, but not yet).
Main constraints:
API could switch infra at some point -> we need to evaluate new-set-up
Big submissions are really big and infrequent -> things sometimes break, diverting attention from current work. Having a rough idea on performance regressions would help prevent that
We would ideally like to know how fast the API can go before something breaks
We should have a series of easy to run load tests, maybe used as an acceptance test before releasing (eventually, but not yet).
Main constraints: