You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Vineet Garg (JIRA)" <ji...@apache.org> on 2017/08/12 22:50:00 UTC

[jira] [Commented] (HIVE-17308) Improvement in join cardinality estimation

    [ https://issues.apache.org/jira/browse/HIVE-17308?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16124738#comment-16124738 ] 

Vineet Garg commented on HIVE-17308:
------------------------------------

First patch introduces different cardinality estimation if there are multiple join keys (since hive doesn't have any way to figure out if there is correlation we always assume correlation).

> Improvement in join cardinality estimation
> ------------------------------------------
>
>                 Key: HIVE-17308
>                 URL: https://issues.apache.org/jira/browse/HIVE-17308
>             Project: Hive
>          Issue Type: Improvement
>          Components: Query Planning
>            Reporter: Vineet Garg
>            Assignee: Vineet Garg
>         Attachments: HIVE-17308.1.patch
>
>
> Currently during logical planning join cardinality is estimated assuming no correlation among join keys (This estimation is done using exponential backoff). Physical planning on the other hand consider correlation for multi keys and uses different estimation. We should consider correlation during logical planning as well.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)