One customer, forty times the size of the rest
You run a shared API for a few hundred businesses. One of them is enormous, most of them are tiny, and all of them are on the same queue.
You cannot ask your largest customer to send less. They are the reason the company has revenue, and their traffic is not abusive — there is simply a lot of it.
Forty requests a second across four customers. Nobody waits more than a second — and that means the smallest customer too, not the average.
Before anything runs: which promise gives way first, and for whom?
The slowest request in a hundred is under 1s
The smallest customers see under 1s too
Every number appears here, on every change.