c21 commented on pull request #35552:
URL: https://github.com/apache/spark/pull/35552#issuecomment-1043249947


   Second to @sunchao and @viirya, would love a more elaborated explanation for 
motivation. btw I think we discussed this before that `HashPartitioning` with 
subset of grouping keys satisfying `ClusteredDistribution`, can be a tradeoff. 
On one hand, it can be bad because it may cause data skew if the cardinality of 
subset of keys are not big enough. On the other hand, it can be good because it 
can help avoid the shuffle before join/group-by/etc. Even if we want to change 
an existing non-stateful operator, or for a new operator to use 
`HashClusteredDistribution` than `ClusteredDistribution`, we still need to 
discuss the tradeoff in depth. 


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to