Start this session to discuss the broker cache implementation.
Some rough idea about this includes:
We can try to cache partial results from composite key<query& segmentList> to corresponding server response. Next query will not hit those already queried segments and only hit segments not in cache.
In request/response to indicate if cache is used or not.
Cache TTL based on segment expiration and also the configurable cache size.
Routing mapping will be online computed.
This is per broker local cache for now, no global cache in picture for now.
Start this session to discuss the broker cache implementation. Some rough idea about this includes: