[IMPALA-3654] Range based pruning for in-predicate

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: Impala 2.6.0
• Fix Version/s: Impala 2.9.0
• Component/s: Frontend
• Labels: parquet performance ramp-up
• Target Version: Product Backlog
• Assignee: Joe McDonnell
• Reporter: Michael Ho
• Created: 31/May/16 21:39
• Updated: 11/May/17 10:57
• Resolved: 08/May/17 22:42

Description

For in-predicates with long in-lists such as TPCDS-Q8, we may benefit by finding the min and max in the list and use them to prune out unqualified scan ranges or filter out rows which don't fall into the range.

Issue Links

• duplicates: IMPALA-4977 Evaluate IN predicates against parquet::Statistics