pan3793 commented on PR #43338: URL: https://github.com/apache/spark/pull/43338#issuecomment-1759050265
Something may be beyond the PR scope: Spark now uses lz4 for shuffle by default, but we switched it to zstd(level 1) internally for over a year, basically gaining the same e2e performance with ~50% shuffle data size reduction. Maybe it's time to migrate the default `spark.io.compression.codec` from lz4 to zstd in Spark 4.0.0 too? Also cc @dongjoon-hyun @wangyum @LuciferYang @mridulm -- 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]
