Proxy Enhancements

[mwyzhao]

• Lower average latency of requests
  • When searching for peers, stop once one is found that meets performance requirements
  • Periodically add and remove well used/unused microservice instances
  • How to handle connections and communication with requester in case of continuous bad network latency

[mwyzhao]

Point 1 (When searching for peers, stop once one is found that meets performance requirements) temporarily put on hold until #3 is finished as it would change the API for searching for peer performances.

Point 2 (Periodically add and remove well used/unused microservice instances) duplicate copies of microservices can be removed from the cache (pick the worst ones maybe). To support this a map needs to be kept to see how many instances of each microservice exists.

Point 3 (How to handle connections and communication with requester in case of continuous bad network latency) No idea how to handle at the moment. Can move this to a separate issue in the future and label it as low priority and worry about it later.

[mwyzhao]

Point 1 is a duplicate of PhysarumSM/common#5 (https://github.com/orgs/Multi-Tier-Cloud/projects/6#card-38822758)

Points 2 and 3 are low priority and have been moved to their own issues. The only downside of not having a mechanism for removing peers is that the cache will become bloated and with the current cache update algorithm, cause a lot of contention. Opened #56. Point 3 is an open question but can be re-opened later when and if we do any stress testing.