[
https://issues.apache.org/jira/browse/CARBONDATA-2528?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16565003#comment-16565003
]
Prasanna Ravichandran commented on CARBONDATA-2528:
---------------------------------------------------
Now the data is fetching from the MV datamap for the order by queries. Working
fine.
explain select attendance,sum(salary)+sum(utilization) as total from
originTable group by attendance order by attendance DESC;
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--+
| plan |
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--+
| == CarbonData Profiler ==
Table Scan on mv_desc_attendance_table
- total blocklets: 4
- filter: none
- pruned by Main DataMap
- skipped blocklets: 0
|
| == Physical Plan ==
*Sort [attendance#12952 DESC NULLS LAST], true, 0
+- Exchange rangepartitioning(attendance#12952 DESC NULLS LAST, 200)
+- *Project [origintable_attendance#12897 AS attendance#12952, total#12898L]
+- *FileScan carbondata
b011.*mv_desc_attendance_table*[origintable_attendance#12897,total#12898L] |
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+–+
explain select empno,sum(salary)+sum(utilization) as total from originTable
group by empno order by empno;
+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--+
| plan |
+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--+
| == CarbonData Profiler ==
Table Scan on mv_order_table
- total blocklets: 6
- filter: none
- pruned by Main DataMap
- skipped blocklets: 0
|
| == Physical Plan ==
*Sort [empno#12822 ASC NULLS FIRST], true, 0
+- Exchange rangepartitioning(empno#12822 ASC NULLS FIRST, 200)
+- *Project [origintable_empno#10724 AS empno#12822, total#10725L]
+- *FileScan carbondata
b011.mv_order_table[origintable_empno#10724,total#10725L] |
+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--+
> MV Datamap - When the MV is created with the order by, then when we execute
> the corresponding query defined in MV with order by, then the data is not
> accessed from the MV.
> ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: CARBONDATA-2528
> URL: https://issues.apache.org/jira/browse/CARBONDATA-2528
> Project: CarbonData
> Issue Type: Bug
> Components: data-query
> Environment: 3 node Opensource ANT cluster. (Opensource Hadoop 2.7.2+
> Opensource Spark 2.2.1+ Opensource Carbondata 1.3.1)
> Reporter: Prasanna Ravichandran
> Assignee: Ravindra Pesala
> Priority: Minor
> Labels: CarbonData, MV, Materialistic_Views
> Fix For: 1.5.0, 1.4.1
>
> Attachments: MV_orderby.docx, data.csv
>
> Time Spent: 6h
> Remaining Estimate: 0h
>
> When the MV is created with the order by condition, then when we execute the
> corresponding query defined in MV along with order by, then the data is not
> accessed from the MV. The data is being accessed from the maintable only.
> Test queries:
> create datamap MV_order using 'mv' as select
> empno,sum(salary)+sum(utilization) as total from originTable group by empno
> order by empno;
> create datamap MV_desc_order using 'mv' as select
> empno,sum(salary)+sum(utilization) as total from originTable group by empno
> order by empno DESC;
> rebuild datamap MV_order;
> rebuild datamap MV_desc_order;
> explain select empno,sum(salary)+sum(utilization) as total from originTable
> group by empno order by empno;
> explain select empno,sum(salary)+sum(utilization) as total from originTable
> group by empno order by empno DESC;
> Expected result: MV with order by condition should access data from the MV
> table only.
>
> Please see the attached document for more details.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
