You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Krisztian Kasa (Jira)" <ji...@apache.org> on 2019/12/20 12:48:00 UTC
[jira] [Updated] (HIVE-22666) Introduce TopNKey operator for PTF
Reduce Sink
[ https://issues.apache.org/jira/browse/HIVE-22666?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Krisztian Kasa updated HIVE-22666:
----------------------------------
Description:
{code}
EXPLAIN EXTENDED
SELECT s_state, ranking
FROM (
SELECT s_state AS s_state,
rank() OVER (PARTITION BY s_state ORDER BY ss_net_profit) AS ranking
FROM testtable_n1000) tmp1
WHERE ranking <= 3;
{code}
In this case the topN value (3) will be pushed to the ReduceSink operator
https://github.com/apache/hive/blob/520aa19b20381bfd2ed25c835443c013f6e6ebb9/ql/src/java/org/apache/hadoop/hive/ql/ppd/OpProcFactory.java#L257
ReduceSink operator uses PTFTopNHash to get the topN rows for each partition key (s_state) value.
The goal of this jira is:
- implement supporting partitioning in TopNKeyOperator
- enable push down of partitioned TopNKeyOperator
> Introduce TopNKey operator for PTF Reduce Sink
> ----------------------------------------------
>
> Key: HIVE-22666
> URL: https://issues.apache.org/jira/browse/HIVE-22666
> Project: Hive
> Issue Type: Improvement
> Reporter: Krisztian Kasa
> Priority: Major
>
> {code}
> EXPLAIN EXTENDED
> SELECT s_state, ranking
> FROM (
> SELECT s_state AS s_state,
> rank() OVER (PARTITION BY s_state ORDER BY ss_net_profit) AS ranking
> FROM testtable_n1000) tmp1
> WHERE ranking <= 3;
> {code}
> In this case the topN value (3) will be pushed to the ReduceSink operator
> https://github.com/apache/hive/blob/520aa19b20381bfd2ed25c835443c013f6e6ebb9/ql/src/java/org/apache/hadoop/hive/ql/ppd/OpProcFactory.java#L257
> ReduceSink operator uses PTFTopNHash to get the topN rows for each partition key (s_state) value.
> The goal of this jira is:
> - implement supporting partitioning in TopNKeyOperator
> - enable push down of partitioned TopNKeyOperator
--
This message was sent by Atlassian Jira
(v8.3.4#803005)