[Performance] Performance inspection over multi-threading, overhead from framework and abstraction  #758

User or operational impact

tried ASAPQuery-backend for network control demo. slightly slower due to no batch query API, he has to send queries again and again; > 8 threads, ASAPQuery-backend is faster than his standalone sketch server.
-> we should understand the threading model of ASAPQuery-backend, and user should be able to control the number of threads for comparison and evaluation
-> when running evaluation/release build, should disable the debugging and info, warn logs output (printing to the stdout will affect query latency), especially when ms level query latency

ASAPQuery-backend is extensible version of standalone server, and code level abstraction adds overhead ?
standalone server stores everything in memory for storing sketches

Metric and target

Representative workload and environment

Current baseline

Correctness constraints

Measurement method and acceptable variance