We have assertions to control the inbound requests. But cannot directly control the number of concurrent connections to a backend server.
I'd like the following features,
1. no limit for inbound requests
2. can set the exact number of outbound concurrent connections (all over the cluster) to protect a backend server
3. the exceed sessions should be able to keep in a queue, not just timeout or reject.
The purpose is to fully use the backend service as well as protecting them.