[
https://issues.apache.org/jira/browse/HIVE-20915?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16713058#comment-16713058
]
Yongzhi Chen commented on HIVE-20915:
-------------------------------------
[~kgyrtkirk], the optimizer uses one reducer to sort the output and put it into
one file (one for each partition, so no need for group by). The behavior is
always like this. You can check old patch(not-checked in) which try to enable
the optimizer by default, enable the optimizer will remove the auto gather
thing.
https://issues.apache.org/jira/secure/attachment/12898098/HIVE-17935.7.patch#file-5
> Make dynamic sort partition optimization available to HoS and MR
> ----------------------------------------------------------------
>
> Key: HIVE-20915
> URL: https://issues.apache.org/jira/browse/HIVE-20915
> Project: Hive
> Issue Type: Bug
> Components: Query Planning
> Affects Versions: 4.0.0
> Reporter: Yongzhi Chen
> Assignee: Yongzhi Chen
> Priority: Major
> Fix For: 4.0.0
>
> Attachments: HIVE-20915.1.patch, HIVE-20915.2.patch,
> HIVE-20915.3.patch, HIVE-20915.4.patch, HIVE-20915.5.patch
>
>
> HIVE-20703 put dynamic sort partition optimization under cost based decision,
> but it also makes the optimizer only available to tez.
> hive.optimize.sort.dynamic.partition works with other execution engines for a
> long time, we should keep the optimizer available to them.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
