You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "L. C. Hsieh (Jira)" <ji...@apache.org> on 2020/08/26 07:02:00 UTC
[jira] [Commented] (SPARK-32361) Remove project if output is subset
of child
[ https://issues.apache.org/jira/browse/SPARK-32361?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17184977#comment-17184977 ]
L. C. Hsieh commented on SPARK-32361:
-------------------------------------
Isn't it already in physical plan phase? Removing such Project in physical plan, seems not giving us too much performance gain.
> Remove project if output is subset of child
> -------------------------------------------
>
> Key: SPARK-32361
> URL: https://issues.apache.org/jira/browse/SPARK-32361
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 3.1.0
> Reporter: ulysses you
> Priority: Minor
>
> We can remove some redundant project after we completed pruning column.
> e.g.,
> {code:java}
> create table t1(c1 int, c2 int) using parquet;
> explain extended
> select sum(c1) from (
> select * from t1
> );
> {code}
> Currently we get this plan.
> {code:java}
> == Physical Plan ==
> *(2) HashAggregate(keys=[], functions=[sum(cast(c1#19 as bigint))], output=[sum(c1)#68L])
> +- Exchange SinglePartition, true, [id=#86]
> +- *(1) HashAggregate(keys=[], functions=[partial_sum(cast(c1#19 as bigint))], output=[sum#70L])
> +- *(1) Project [c1#19]
> +- *(1) ColumnarToRow
> +- FileScan parquet default.t1[c1#19] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex[hdfs:///user/hive/warehouse/t1], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<c1:int>
> {code}
> We can remove the `Project`, like this
> {code:java}
> == Physical Plan ==
> *(2) HashAggregate(keys=[], functions=[sum(cast(c1#19 as bigint))], output=[sum(c1)#68L])
> +- Exchange SinglePartition, true, [id=#86]
> +- *(1) HashAggregate(keys=[], functions=[partial_sum(cast(c1#19 as bigint))], output=[sum#70L])
> +- *(1) ColumnarToRow
> +- FileScan parquet default.t1[c1#19] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex[hdfs:///user/hive/warehouse/t1], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<c1:int>
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org