Question on result level caching on brokers - even for a cluster that has data being written to via both batch and realtime ingestion, is there a reason why we should not have the result level cache on by default? For queries against realtime segments or the active time period for which data is being ingested, result level cache will be invalidated. Does that cache miss actually result in a big perf hit? Has any one tried turning it on? We are seeing great results with query result caching on for brokers. So I am tempted to turn it on across the board. CC: @Gian Merlino - I would be curious to know more your thought on this since we talked about caching a bit during the recent meet up.