[jira] [Commented] (DRILL-4706) Fragment planning causes Drillbits to read remote chunks when local copies are available

["Jacques Nadeau (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-4706?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15314839#comment-15314839 ] 

Jacques Nadeau commented on DRILL-4706:
---------------------------------------

Drill tries to balance parallelism and distribution. The factor that influences this is called planner.affinity_factor. It doesn't guarantee locality but varying it will impact the balance. HDFS does not currently implement a required locality concept.

> Fragment planning causes Drillbits to read remote chunks when local copies are available
> ----------------------------------------------------------------------------------------
>
>                 Key: DRILL-4706
>                 URL: https://issues.apache.org/jira/browse/DRILL-4706
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Query Planning & Optimization
>    Affects Versions: 1.6.0
>         Environment: CentOS, RHEL
>            Reporter: Kunal Khatua
>              Labels: performance, planning
>
> When a table (datasize=70GB) of 160 parquet files (each having a single rowgroup and fitting within one chunk) is available on a 10-node setup with replication=3 ; a pure data scan query causes about 2% of the data to be read remotely. 
> Even with the creation of metadata cache, the planner is selecting a sub-optimal plan of executing the SCAN fragments such that some of the data is served from a remote server. 



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