Provide a mechanism to smooth out large surges in request volume:
Persist request documents and metatdata as usual.
If request volume exceeds a specified threshold (requests/second), instead of invoking from the listener right away, queue the request IDs in a persistent store for later processing.
Drain requests according to their IDs at a designated rate.
Request queuing and throttling should be optional based on configuration.
Provide a mechanism to smooth out large surges in request volume:
Request queuing and throttling should be optional based on configuration.