In a period where microservices become more and more predominant, I was wondering if there are some common techniques by which Java Web Services detect overload before start deteriorating.
Unfortunately, it doesn't look like there is One golden rule (e.g. CPU never has to go above 50%), and it seems a topic where the best heuristic wins.
Online there are several resources pointing out to a X number of CPU, GC pauses, Network IO etc, but I was wondering if there were more scientific studies/documented use cases to catch these signaling.