hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jesus Camacho Rodriguez (Jira)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-23298) Disable RS deduplication step in Optimizer if it is run in TezCompiler
Date Sun, 03 May 2020 19:11:00 GMT

    [ https://issues.apache.org/jira/browse/HIVE-23298?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17098541#comment-17098541
] 

Jesus Camacho Rodriguez commented on HIVE-23298:
------------------------------------------------

Thanks [~ashutoshc]. Changes LGTM, +1 (pending tests)

Do you have any theory for the number of rows / data size changing (most of the times to the
half)? It should not be related to this patch per se but probably worth exploring at some
point... Maybe it is a side effect of running annotation of stats before vs after dedup, i.e.,
two RS after each other is accounting for double reduction?

> Disable RS deduplication step in Optimizer if it is run in TezCompiler
> ----------------------------------------------------------------------
>
>                 Key: HIVE-23298
>                 URL: https://issues.apache.org/jira/browse/HIVE-23298
>             Project: Hive
>          Issue Type: Improvement
>          Components: Physical Optimizer
>            Reporter: Jesus Camacho Rodriguez
>            Assignee: Jesus Camacho Rodriguez
>            Priority: Major
>         Attachments: HIVE-23298.1.patch, HIVE-23298.2.patch, HIVE-23298.patch
>
>
> HIVE-20703 introduced an additional RS deduplication step in TezCompiler. We could possibly
try to disable the one that runs in {{Optimizer}} if we are using Tez so we do not run the
optimization twice.
> This issue is to explore that possibility.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message