[
https://issues.apache.org/jira/browse/DRILL-4528?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Krystal updated DRILL-4528:
---------------------------
Labels: limit0 (was: )
> AVG() windows function is not optimized for limit 0 queries
> -----------------------------------------------------------
>
> Key: DRILL-4528
> URL: https://issues.apache.org/jira/browse/DRILL-4528
> Project: Apache Drill
> Issue Type: Bug
> Components: Query Planning & Optimization
> Reporter: Krystal
> Assignee: Sean Hsuan-Yi Chu
> Labels: limit0
>
> git.commit.id.abbrev=cee5317
> The following sample query contains the avg() windows function that is not
> optimized when wrapped with limit 0:
> select * from (
> SELECT AVG(cast( col1 as BIGINT )) OVER(PARTITION BY cast( col4 as TIMESTAMP)
> ORDER BY cast( col5 as DATE )) FROM `fewRowsAllData_v`) t limit 0
> Physical Plan:
> {code}
> 00-00 Screen : rowType = RecordType(ANY EXPR$0): rowcount = 1.0,
> cumulative cost = {469.1 rows, 5717.190984570043 cpu, 0.0 io, 0.0 network,
> 1872.0 memory}, id = 5762023
> 00-01 Project(EXPR$0=[$0]) : rowType = RecordType(ANY EXPR$0): rowcount
> = 1.0, cumulative cost = {469.0 rows, 5717.090984570043 cpu, 0.0 io, 0.0
> network, 1872.0 memory}, id = 5762022
> 00-02 SelectionVectorRemover : rowType = RecordType(ANY EXPR$0):
> rowcount = 1.0, cumulative cost = {469.0 rows, 5717.090984570043 cpu, 0.0 io,
> 0.0 network, 1872.0 memory}, id = 5762021
> 00-03 Limit(fetch=[0]) : rowType = RecordType(ANY EXPR$0): rowcount
> = 1.0, cumulative cost = {468.0 rows, 5716.090984570043 cpu, 0.0 io, 0.0
> network, 1872.0 memory}, id = 5762020
> 00-04 Project(EXPR$0=[/(CastHigh($3), $4)]) : rowType =
> RecordType(ANY EXPR$0): rowcount = 78.0, cumulative cost = {468.0 rows,
> 5716.090984570043 cpu, 0.0 io, 0.0 network, 1872.0 memory}, id = 5762019
> 00-05 Window(window#0=[window(partition {2} order by [1] range
> between UNBOUNDED PRECEDING and CURRENT ROW aggs [SUM($0), COUNT($0)])]) :
> rowType = RecordType(BIGINT $0, DATE $1, TIMESTAMP(0) $2, BIGINT w0$o0,
> BIGINT w0$o1): rowcount = 78.0, cumulative cost = {390.0 rows,
> 5404.090984570043 cpu, 0.0 io, 0.0 network, 1872.0 memory}, id = 5762018
> 00-06 SelectionVectorRemover : rowType = RecordType(BIGINT $0,
> DATE $1, TIMESTAMP(0) $2): rowcount = 78.0, cumulative cost = {312.0 rows,
> 5170.090984570043 cpu, 0.0 io, 0.0 network, 1872.0 memory}, id = 5762017
> 00-07 Sort(sort0=[$2], sort1=[$1], dir0=[ASC], dir1=[ASC]) :
> rowType = RecordType(BIGINT $0, DATE $1, TIMESTAMP(0) $2): rowcount = 78.0,
> cumulative cost = {234.0 rows, 5092.090984570043 cpu, 0.0 io, 0.0 network,
> 1872.0 memory}, id = 5762016
> 00-08 Project($0=[CAST(CAST($0):BIGINT):BIGINT],
> $1=[CAST(CAST($1):DATE):DATE], $2=[CAST(CAST($2):TIMESTAMP(0)):TIMESTAMP(0)])
> : rowType = RecordType(BIGINT $0, DATE $1, TIMESTAMP(0) $2): rowcount = 78.0,
> cumulative cost = {156.0 rows, 1170.0 cpu, 0.0 io, 0.0 network, 0.0 memory},
> id = 5762015
> 00-09 Scan(groupscan=[ParquetGroupScan
> [entries=[ReadEntryWithPath
> [path=maprfs:///drill/testdata/window_functions/fewRowsAllData.parquet]],
> selectionRoot=maprfs:/drill/testdata/window_functions/fewRowsAllData.parquet,
> numFiles=1, usedMetadataFile=false, columns=[`col1`, `col5`, `col4`]]]) :
> rowType = RecordType(ANY col1, ANY col5, ANY col4): rowcount = 78.0,
> cumulative cost = {78.0 rows, 234.0 cpu, 0.0 io, 0.0 network, 0.0 memory}, id
> = 5762014
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
