[jira] [Commented] (HIVE-9780) Add another level of explain for RDBMS audience

["Carter Shanklin (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-9780?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15007080#comment-15007080 ] 

Carter Shanklin commented on HIVE-9780:
---------------------------------------

This is so much better than the old style explain plans we should consider making it the default for Hive 2.0.

> Add another level of explain for RDBMS audience
> -----------------------------------------------
>
>                 Key: HIVE-9780
>                 URL: https://issues.apache.org/jira/browse/HIVE-9780
>             Project: Hive
>          Issue Type: New Feature
>          Components: Diagnosability
>            Reporter: Pengcheng Xiong
>            Assignee: Pengcheng Xiong
>            Priority: Minor
>             Fix For: 1.2.0
>
>         Attachments: HIVE-9780.01.patch, HIVE-9780.02.patch, HIVE-9780.03.patch, HIVE-9780.04.patch, HIVE-9780.05.patch, HIVE-9780.06.patch, HIVE-9780.07.patch, HIVE-9780.08.patch, HIVE-9780.09.patch, HIVE-9780.10.patch
>
>
> Current Hive Explain (default) is targeted at MR Audience. We need a new level of explain plan to be targeted at RDBMS audience. The explain requires these:
> 1) The focus needs to be on what part of the query is being executed rather than internals of the engines
> 2) There needs to be a clearly readable tree of operations
> 3) Examples - Table scan should mention the table being scanned, the Sarg, the size of table and expected cardinality after the Sarg'ed read. The join should mention the table being joined with and the join condition. The aggregate should mention the columns in the group-by. 



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