Rui Guo created ASTERIXDB-2837:
----------------------------------
Summary: Unexpected long query runtime
Key: ASTERIXDB-2837
URL: https://issues.apache.org/jira/browse/ASTERIXDB-2837
Project: Apache AsterixDB
Issue Type: Bug
Reporter: Rui Guo
Attachments: Original_query_plan.txt, Splitted_query_plan.txt
A simple scan-and-filter query takes much longer than expected to run.
The query is at
asterixdb/asterix-app/src/test/resources/runtimets/queries_sqlpp/aggregate-subclause/agg_filter_01/agg_filter_01.3.query.sqlpp,
and the SQLPP query is
{code:java}
select
count(*) filter(where four > 0) as cnt,
sum(four) filter(where four > 0) as sm
from tenk;{code}
The above query takes around *half a minute* to execute. However, if we split
it to two selects, say,
{code:java}
select
sum(four) filter(where four > 0) as sm
from tenk;
select
count(*) filter(where four > 0) as cnt
from tenk;
{code}
then the runtime would be less than *0.1 second*.
The related optimized query plans are attached.
And the dataset load statement is at
asterixdb/asterix-app/src/test/resources/runtimets/queries_sqlpp/aggregate-subclause/agg_filter_01/agg_filter_01.2.update.sqlpp
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
