Query peroformance issue

Hi Hello ,
We are using memcached for caching , we have enable caching for historical node/broker.
It works well with topN small range query query like a day or 2 days . However , when we have multiple query (15-20) with little longer range say a week it goes beyond 3-4 sec for each query , collectively for 15 query in one dashboard it really takes visibly longer time to load . Our data is segmented with hour granularity and avg size of segment is 500-600 mb .
Both broker/Historical machine is r3.8x
Historical config
druid.historical.cache.populateCache=true
druid.processing.buffer.sizeBytes=1073741824
druid.processing.numThreads=31
druid.segmentCache.locations=[{“path”: “/mnt/persistent/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt/persistent2/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt/persistent3/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt/persistent4/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt/persistent5/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt/persistent6/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt2/persistent/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt2/persistent2/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt2/persistent3/zk_druid”, “maxSize”: 300000000000},{“path”: “/mnt2/persistent4/zk_druid”, “maxSize”: 300000000000}]
druid.server.http.numThreads=50
druid.server.maxSize=2998000000000

Broker config

Correct me if I’m wrong In the document it mentions only config cache either historical or broker node but not both right?

You can have the cache on both but generally it doesn’t make sense to do so.

What is required to diagnose slow running queries is query metrics from http://druid.io/docs/0.9.1.1/operations/metrics.html