[HIVE-11428] Performance: Struct IN() clauses are extremely slow (~10x slower)

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.3.0, 2.0.0
• Fix Version/s: 1.3.0, 2.0.0
• Component/s: Query Processor
• Assignee: Hari Sankar Sivarama Subramaniyan
• Reporter: Gopal Vijayaraghavan
• Created: 31/Jul/15 21:36
• Updated: 16/Feb/16 23:52
• Resolved: 07/Aug/15 21:19

Description

Hive today does not support tuple IN() clauses today, but provides a way to rewrite (a,b) IN (...) using complex types.

select * from table where STRUCT(a,b) IN (STRUCT(1,2), STRUCT(2,3) ...);

This would be fine, except it is massively slower due to ObjectConvertors and Struct constructor not being constant folded.

Issue Links

• is depended upon by: HIVE-11468 Vectorize: Struct IN() clauses
• is part of: HIVE-11315 Compilation performance issues for large queries
• is required by: HIVE-11462 GenericUDFStruct should constant fold at compile time