[ https://issues.apache.org/jira/browse/HIVE-1544?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12899096#action_12899096 ]
Ning Zhang commented on HIVE-1544: ---------------------------------- The JoinDesc already has a flag noOuterJoin to keep track if there are outer joins involved in the join operator. Based on that we should set a flag in the ReduceSinkDesc to indicate whether NULL-keyed rows will be filtered out. > Filtering out NULL-keyed rows in ReduceSinkOperator when no outer join > involved > ------------------------------------------------------------------------------- > > Key: HIVE-1544 > URL: https://issues.apache.org/jira/browse/HIVE-1544 > Project: Hadoop Hive > Issue Type: Improvement > Reporter: Ning Zhang > > As discussed in HIVE-741, if a plan indicates that a non-outer join is the > first operator in the reducer, the ReduceSinkOperator should filter out (not > sending) rows with NULL as keys since they will not generate any results > anyways. This should save both bandwidth and processing power. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.