Feature: Load Shedding

[jameskleeh]

Feature description

Provide a solution integrated into Micronaut to support load shedding. There is a seemingly unmaintained library from Netflix https://github.com/Netflix/concurrency-limits that has implementations of common patterns in this space. Functionality that Micronaut could provide include:

1. Annotations to indicate the priority of an API. Requests with lower priority would be shed first
2. Integration into the server with one or more algorithms to shed load based on several factors adjustable by the user

• Some applications may be memory bound, cpu bound, network bound, etc...

The end goal here is a feature that could be demoed to show an ever increasing load sent to the server causes it to reject requests while maintaining the latency for the requests it does process. After a sustained period of traffic beyond the capabilities of the server the traffic would drop to a normal level and observe the server recover in a short amount of time.