Details
-
Sub-task
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
Description
Currently, for sort-shuffle the generation of subpartition data storage orderĀ is random shuffle. However, if there is no enough resources to run the downstream consumer tasks in parallel, the performance can be influenced because of the random disk IO caused by the random subpartition data storage order. This ticket aims to improve this scenario.
Attachments
Issue Links
- links to