Yeah, strategically place queues in your architecture, artificially limit size of queues in stress tests (like how you can use rlimit for other resources), and track the derivative of queue fill.
Reject new entries when the queue is full - all the way back down the line to whatever wanted to do the work in the first place, such as a user.
From 2019, but still relevant.
Still no update of the book he is writing, though.