Mailing-List: contact issues-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Date: Mon, 24 Jul 2017 08:19:02 +0000 (UTC)
From: "liyunzhang_intel (JIRA)" <jira@apache.org>
To: issues@hive.apache.org
Message-ID: <JIRA.13087017.1499981578000.313419.1500884342760@Atlassian.JIRA>
In-Reply-To: <JIRA.13087017.1499981578000@Atlassian.JIRA>
References: <JIRA.13087017.1499981578000@Atlassian.JIRA> <JIRA.13087017.1499981578989@jira-lw-us.apache.org>
Subject: [jira] [Commented] (HIVE-17087) Remove unnecessary HoS DPP trees
 during map-join conversion
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Mon, 24 Jul 2017 08:19:11 -0000


    [ https://issues.apache.org/jira/browse/HIVE-17087?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16098064#comment-16098064 ] 

liyunzhang_intel commented on HIVE-17087:
-----------------------------------------

[~stakiar]
{quote}
  It's actually suppose to be SparkMapJoinOptimizer and SparkRemoveDynamicPruningBySize. I think the ordering may not be deterministic because both are applied using the the same GraphWalker. 
{quote}
  
opRules is LinkedHashMap which is insertion-ordered and why SparkMapJoinOptimizer maybe executed later than SparkRemoveDynamicPruningBySize?

> Remove unnecessary HoS DPP trees during map-join conversion
> -----------------------------------------------------------
>
>                 Key: HIVE-17087
>                 URL: https://issues.apache.org/jira/browse/HIVE-17087
>             Project: Hive
>          Issue Type: Sub-task
>          Components: Spark
>            Reporter: Sahil Takiar
>            Assignee: Sahil Takiar
>         Attachments: HIVE-17087.1.patch, HIVE-17087.2.patch
>
>
> Ran the following query in the {{TestSparkCliDriver}}:
> {code:sql}
> set hive.spark.dynamic.partition.pruning=true;
> set hive.auto.convert.join=true;
> create table partitioned_table1 (col int) partitioned by (part_col int);
> create table partitioned_table2 (col int) partitioned by (part_col int);
> create table regular_table (col int);
> insert into table regular_table values (1);
> alter table partitioned_table1 add partition (part_col = 1);
> insert into table partitioned_table1 partition (part_col = 1) values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10);
> alter table partitioned_table2 add partition (part_col = 1);
> insert into table partitioned_table2 partition (part_col = 1) values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10);
> explain select * from partitioned_table1, partitioned_table2 where partitioned_table1.part_col = partitioned_table2.part_col;
> {code}
> and got the following explain plan:
> {code}
> STAGE DEPENDENCIES:
>   Stage-2 is a root stage
>   Stage-3 depends on stages: Stage-2
>   Stage-1 depends on stages: Stage-3
>   Stage-0 depends on stages: Stage-1
> STAGE PLANS:
>   Stage: Stage-2
>     Spark
> #### A masked pattern was here ####
>       Vertices:
>         Map 3 
>             Map Operator Tree:
>                 TableScan
>                   alias: partitioned_table1
>                   Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                   Select Operator
>                     expressions: col (type: int), part_col (type: int)
>                     outputColumnNames: _col0, _col1
>                     Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                     Select Operator
>                       expressions: _col1 (type: int)
>                       outputColumnNames: _col0
>                       Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                       Group By Operator
>                         keys: _col0 (type: int)
>                         mode: hash
>                         outputColumnNames: _col0
>                         Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                         Spark Partition Pruning Sink Operator
>                           partition key expr: part_col
>                           Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                           target column name: part_col
>                           target work: Map 2
>   Stage: Stage-3
>     Spark
> #### A masked pattern was here ####
>       Vertices:
>         Map 2 
>             Map Operator Tree:
>                 TableScan
>                   alias: partitioned_table2
>                   Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                   Select Operator
>                     expressions: col (type: int), part_col (type: int)
>                     outputColumnNames: _col0, _col1
>                     Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                     Spark HashTable Sink Operator
>                       keys:
>                         0 _col1 (type: int)
>                         1 _col1 (type: int)
>             Local Work:
>               Map Reduce Local Work
>   Stage: Stage-1
>     Spark
> #### A masked pattern was here ####
>       Vertices:
>         Map 1 
>             Map Operator Tree:
>                 TableScan
>                   alias: partitioned_table1
>                   Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                   Select Operator
>                     expressions: col (type: int), part_col (type: int)
>                     outputColumnNames: _col0, _col1
>                     Statistics: Num rows: 10 Data size: 11 Basic stats: COMPLETE Column stats: NONE
>                     Map Join Operator
>                       condition map:
>                            Inner Join 0 to 1
>                       keys:
>                         0 _col1 (type: int)
>                         1 _col1 (type: int)
>                       outputColumnNames: _col0, _col1, _col2, _col3
>                       input vertices:
>                         1 Map 2
>                       Statistics: Num rows: 11 Data size: 12 Basic stats: COMPLETE Column stats: NONE
>                       File Output Operator
>                         compressed: false
>                         Statistics: Num rows: 11 Data size: 12 Basic stats: COMPLETE Column stats: NONE
>                         table:
>                             input format: org.apache.hadoop.mapred.SequenceFileInputFormat
>                             output format: org.apache.hadoop.hive.ql.io.HiveSequenceFileOutputFormat
>                             serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
>             Local Work:
>               Map Reduce Local Work
>   Stage: Stage-0
>     Fetch Operator
>       limit: -1
>       Processor Tree:
>         ListSink
> {code}
> Stage-2 seems unnecessary, given that Stage-1 is going to do a full table scan of {{partitioned_table1}} when running the map-join


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