Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
Problem:When opening a input stream we are creating client every time and each time it is taking around 400 ms, In case of detail query we are opening for each blocklet and it is impacting the query performance
Solution:As query execution is a reading operation we can open only one client
Impact Area:Query execution and compaction
Need to handle for Below scenario
1. All inputstream creation
2. All outputstream creation
3. All places where we are creating carbon file(HDFS)
In this issue we are handling scenario1