hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Dere (JIRA)" <>
Subject [jira] [Updated] (HIVE-16154) Determine when dynamic runtime filtering should be disabled
Date Wed, 22 Mar 2017 06:49:41 GMT


Jason Dere updated HIVE-16154:
    Attachment: HIVE-16154.2.patch

Few changes to the patch:
- Do cost/benefit check of semijoin optimizations first, and cycle detection last
- Change cost/benefit to be in terms of rows, per [~gopalv]'s suggestion
- Compute benefit in terms of the FilterOperator following the TableScan, rather than the
- Fix findColumnOrigin to work in the case of multiple parents

> Determine when dynamic runtime filtering should be disabled
> -----------------------------------------------------------
>                 Key: HIVE-16154
>                 URL:
>             Project: Hive
>          Issue Type: Bug
>          Components: Query Planning
>            Reporter: Jason Dere
>            Assignee: Jason Dere
>         Attachments: HIVE-16154.1.patch, HIVE-16154.2.patch
> Currently dynamic min/max/bloom optimization is always enabled. However there are times
where it may not be beneficial, such as if the semijoin has a PK-FK relation and there are
no filters on the semijoin table. Try to devise a way to do a cost/benefit calculation to
see if there is enough benefit to adding the runtime filter.

This message was sent by Atlassian JIRA

View raw message