junlong created SPARK-3917: ------------------------------ Summary: Compress data before network transfer Key: SPARK-3917 URL: https://issues.apache.org/jira/browse/SPARK-3917 Project: Spark Issue Type: Improvement Components: Spark Core Affects Versions: 1.1.0 Environment: All Reporter: junlong Priority: Critical Fix For: 1.1.0
When training Gradient Boosting Decision Tree on large sparse data, heavy network flow pull down CPU utilization ratio. And through compression on network flow data, 90% are reduced. So maybe compression before transfering may provide higher speedup on spark. And user can configure it to decide whether compress or not. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org