[HIVE-17877] HoS: combine equivalent DPP sink works

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.0.0
• Component/s: Spark
• Assignee: Rui Li
• Reporter: Rui Li
• Created: 23/Oct/17 14:33
• Updated: 22/May/18 23:17
• Resolved: 06/Nov/17 02:26

Description

Suppose part1 and part2 are partitioned tables. The simplest use case should be something like:

explain
select * from
  (select part1.key, part1.value from part1 join src on part1.p=src.key) a
union all
  (select part2.key, part2.value from part2 join src on part2.p=src.key);

Issue Links

• breaks: HIVE-18111 Fix temp path for Spark DPP sink
• is required by: HIVE-17193 HoS: don't combine map works that are targets of different DPPs
• links to: RB link