cloud-fan commented on issue #25295: [SPARK-28560][SQL] Optimize shuffle reader to local shuffle reader when smj converted to bhj in adaptive execution URL: https://github.com/apache/spark/pull/25295#issuecomment-530372266 I think this is a good idea, but the implementation needs more polishing. What I expect to see is: 1. a rule to convert `ShuffleQueryStageExec` to `LocalShuffleReaderExec`. This rule must be run before `ReduceNumShufflePartitions`. 2. A special `ShuffledRDD` that reads shuffle blocks grouped by mapId instead of reduceId. I'm a little worried to see the invasive changes in the underlying shuffle component. Can you briefly explain how your special `ShuffledRDD` is implemented?
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
