kamal Maheshwari created IMPALA-6534:
----------------------------------------
Summary: Impala query planner should be smart to remove extra
GROUP BY
Key: IMPALA-6534
URL: https://issues.apache.org/jira/browse/IMPALA-6534
Project: IMPALA
Issue Type: Improvement
Components: Backend
Affects Versions: Impala 2.11.0
Reporter: kamal Maheshwari
Attachments: Query_cte_explain_plan.txt, Query_cte_profile.txt,
Query_no_cte_explain_plan.txt, Query_no_cte_profile.txt, query_cte.hql,
query_no_cte.hql
A customer shared a complex query (attached) from their existing DB environment
which uses CTE. They noticed that the query didn't finish as-is. They took out
the CTE block and the query finished quickly.
I then suggested to take out the GROUP BY from the CTE block and the query
finished fairly quickly.
The ask is - if the planner can detect such situations and do the right thing.
These are common scenarios for workload migrations.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
