spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Apache Spark (JIRA)" <j...@apache.org>
Subject [jira] [Assigned] (SPARK-21707) Improvement a special case for non-deterministic filters in optimizer
Date Tue, 29 Aug 2017 13:25:00 GMT

     [ https://issues.apache.org/jira/browse/SPARK-21707?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Apache Spark reassigned SPARK-21707:
------------------------------------

    Assignee:     (was: Apache Spark)

> Improvement a special case for non-deterministic filters in optimizer
> ---------------------------------------------------------------------
>
>                 Key: SPARK-21707
>                 URL: https://issues.apache.org/jira/browse/SPARK-21707
>             Project: Spark
>          Issue Type: Improvement
>          Components: SQL
>    Affects Versions: 2.3.0
>            Reporter: caoxuewen
>
> Currently, Did a lot of special handling for non-deterministic projects and filters in
optimizer. but not good enough. this patch add a new special case for non-deterministic filters.
Deal with that we only need to read user needs fields for non-deterministic filters in optimizer.
> For example, the condition of filters is nondeterministic. e.g:contains nondeterministic
function(rand function), HiveTableScans optimizer generated:
> ```
> HiveTableScans plan:Aggregate [k#2L], [k#2L, k#2L, sum(cast(id#1 as bigint)) AS sum(id)#395L]
> +- Project [d004#205 AS id#1, CEIL(c010#214) AS k#2L]
>    +- Filter ((isnotnull(d004#205) && (rand(-4530215890880734772) <= 0.5))
&& NOT (cast(cast(d004#205 as decimal(10,0)) as decimal(11,1)) = 0.0))
>       +- MetastoreRelation XXX_database, XXX_table
> HiveTableScans plan:Project [d004#205 AS id#1, CEIL(c010#214) AS k#2L]
> +- Filter ((isnotnull(d004#205) && (rand(-4530215890880734772) <= 0.5)) &&
NOT (cast(cast(d004#205 as decimal(10,0)) as decimal(11,1)) = 0.0))
>    +- MetastoreRelation XXX_database, XXX_table
> HiveTableScans plan:Filter ((isnotnull(d004#205) && (rand(-4530215890880734772)
<= 0.5)) && NOT (cast(cast(d004#205 as decimal(10,0)) as decimal(11,1)) = 0.0))
> +- MetastoreRelation XXX_database, XXX_table
> HiveTableScans plan:MetastoreRelation XXX_database, XXX_table
> ```
> so HiveTableScan will read all the fields from table. but we only need to ‘d004’
and 'c010' . it will affect the performance of task.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message