This is exactly the kind of scenario where numbers behave perfectly fine on paper, but reality is hiding in the space between the variables. And it’s such a classic trap in performance analytics—everything looks “within limits” until you start exploring how two seemingly normal metrics behave together.

Averages, percentiles, even distribution curves—they’re all comforting because they suggest stability. But they also flatten nuance. They strip away the messy interactions that real-world systems thrive on (or break because of).
So you get this odd situation where every metric looks healthy individually, yet users are still complaining. And it’s only when you put processing time and resource consumption side-by-side that the story starts to reveal itself.