Steve Carlin created IMPALA-14995:
-------------------------------------
Summary: Optimization for count star with partitioned group
Key: IMPALA-14995
URL: https://issues.apache.org/jira/browse/IMPALA-14995
Project: IMPALA
Issue Type: Improvement
Reporter: Steve Carlin
In the test file {{iceberg-plain-count-star-optimization.test for the following
table:}}
{{}}
{code:java}
create table ice_tbl (col_i INT, col_s STRING ) partitioned by spec (col_s)
stored as iceberg tblproperties ('write.format.default' = 'parquet');{code}
{{The following test exists which could be using count star optimization, but
currently is not:}}
{{}}
{code:java}
---- QUERY
select
count(*)
from ice_tbl
group by col_s;
---- RESULTS
4
4
---- RUNTIME_PROFILE
aggregation(SUM, NumRowGroups): 4
aggregation(SUM, NumFileMetadataRead): 0{code}
When the count star optimization gets implemented, the NumRowGroups should show
0 and NumFileMetadataRes should show 4.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
