[
https://issues.apache.org/jira/browse/SPARK-21436?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16558013#comment-16558013
]
zhengruifeng edited comment on SPARK-21436 at 7/26/18 7:37 AM:
---------------------------------------------------------------
[~holdenk] It looks like that \{{distinct}} already utilized the known
partitioner. \{distinct} calls
\{{color:#ffc66d}{color:#333333}combineByKeyWithClassTag} internally, and will
avoid the shuffle if rdd's partitioner is equal to a hash partitioner with same
#partitions.{color}
Or do you mean that we need to expose some API like \{distinct(partitioner:
Partitioner)} for other kinds of partitioners like \{RangePartitioner}?
was (Author: podongfeng):
[~holdenk] It looks like that \{distinct} already utilized the known
partitioner. \{distinct} calls
{{color:#ffc66d}{color:#333333}combineByKeyWithClassTag} internally, and will
avoid the shuffle if rdd's partitioner is equal to a hash partitioner with same
#partitions.
{color}{color}
{color:#ffc66d}{color:#333333}Or do you mean that we need to expose some API
like {distinct(partitioner: {color}{color}Partitioner)} for other kinds of
partitioners like \{RangePartitioner}?
> Take advantage of known partioner for distinct on RDDs
> ------------------------------------------------------
>
> Key: SPARK-21436
> URL: https://issues.apache.org/jira/browse/SPARK-21436
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core
> Affects Versions: 2.3.0
> Reporter: holdenk
> Priority: Minor
>
> If we have a known partitioner we should be able to avoid the shuffle.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
