hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Na Yang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-8054) Disable hive.optimize.union.remove when hive.execution.engine=spark [Spark Branch]
Date Mon, 15 Sep 2014 23:07:34 GMT

    [ https://issues.apache.org/jira/browse/HIVE-8054?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14134668#comment-14134668
] 

Na Yang commented on HIVE-8054:
-------------------------------

Hi [~xuefuz], yes. both load_dyn_part13 and optimize_nullscan are related to my patch. I regenerated
the .q.out files for both of them. Thanks.

> Disable hive.optimize.union.remove when hive.execution.engine=spark [Spark Branch]
> ----------------------------------------------------------------------------------
>
>                 Key: HIVE-8054
>                 URL: https://issues.apache.org/jira/browse/HIVE-8054
>             Project: Hive
>          Issue Type: Improvement
>          Components: Spark
>            Reporter: Xuefu Zhang
>            Assignee: Na Yang
>              Labels: Spark-M1
>         Attachments: HIVE-8054-spark.patch, HIVE-8054.2-spark.patch
>
>
> Option hive.optimize.union.remove introduced in HIVE-3276 removes union operators from
the operator graph in certain cases as an optimization reduce the number of MR jobs. While
making sense in MR, this optimization is actually harmful to an execution engine such as Spark,
which natives supports union without requiring additional jobs. This is because removing union
operator creates disjointed operator graphs, each graph generating a job, and thus this optimization
requires more jobs to run the query. Not to mention the additional complexity handling linked
FS descriptors.
> I propose that we disable such optimization when the execution engine is Spark.



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

Mime
View raw message