Re-use HBee cache
rdettai opened this issue · comments
Work is distributed randomly to HBees because the query is transferred upon invocation. This means that subsequent queries have little probability to attribute the workloads with possible cache hits (e.g on the data downloaded from S3).
A possible solution would be to let the HBees first connect to the HComb, registering the state of their cache, which would then be able to assign work according to this state.