We really need some decent way of testing end-end performance improvements like this claim.
It is impractical to rely on manual testing to detect regressions. And a bunch of unit tests for the throttling logic does not adequately prove the real world behaviour of the system.
*Needs Discussion*
We really need some decent way of testing end-end performance improvements like this claim.
It is impractical to rely on manual testing to detect regressions. And a bunch of unit tests for the throttling logic does not adequately prove the real world behaviour of the system.