You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@spark.apache.org by "David Sabater (JIRA)" <ji...@apache.org> on 2015/07/29 13:51:04 UTC

[jira] [Commented] (SPARK-3862) MultiWayBroadcastInnerHashJoin

    [ https://issues.apache.org/jira/browse/SPARK-3862?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14645915#comment-14645915 ] 

David Sabater commented on SPARK-3862:
--------------------------------------

This may sound too extreme but it will be great to have an option in SparkSQL to broadcast these dimension tables before even actually run the queries, which I think will speed up the actual query execution massively (Other SQL MPP engines are doing that already).
It will be a call similar to CACHE but replicating all partitions accross all nodes.

> MultiWayBroadcastInnerHashJoin
> ------------------------------
>
>                 Key: SPARK-3862
>                 URL: https://issues.apache.org/jira/browse/SPARK-3862
>             Project: Spark
>          Issue Type: Sub-task
>          Components: SQL
>            Reporter: Reynold Xin
>
> It is common to have a single fact table inner join many small dimension tables.  We can exploit this fact and create a MultiWayBroadcastInnerHashJoin (or maybe just MultiwayDimensionJoin) operator that optimizes for this pattern.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org