[ https://issues.apache.org/jira/browse/SPARK-13353?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15176598#comment-15176598 ]
Davies Liu commented on SPARK-13353: ------------------------------------ [~rxin] Any idea on this one? One workaround could be packing rows as batches before collect a DataFrame, then deserialize them in collect() on driver. > Use UnsafeRowSerializer to collect DataFrame > -------------------------------------------- > > Key: SPARK-13353 > URL: https://issues.apache.org/jira/browse/SPARK-13353 > Project: Spark > Issue Type: Improvement > Components: SQL > Reporter: Davies Liu > > UnsafeRowSerializer should be more efficient than JavaSerializer or > KyroSerializer for DataFrame. -- 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