[IMPALA-820] Missed pushdown opportunity when HAVING predicates refer to non-aggregates

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 1.2.3
• Fix Version/s: Impala 1.2.4
• Assignee: Unassigned
• Reporter: Henry Robinson
• Created: 19/Feb/14 18:24
• Updated: 02/Mar/14 01:46
• Resolved: 02/Mar/14 01:46

Description

If a HAVING predicate refers only to a column in the grouped table (rather than an aggregate), it can be pushed down past the aggregation:

[localhost:21000] > explain select bool_col from functional.alltypes group by bool_col having bool_col = false;
Query: explain select bool_col from functional.alltypes group by bool_col having bool_col = false
+-----------------------------------------------------------+
| Explain String                                            |
+-----------------------------------------------------------+
| Estimated Per-Host Requirements: Memory=170.00MB VCores=2 |
|                                                           |
| 04:EXCHANGE [PARTITION=UNPARTITIONED]                     |
| |                                                         |
| 03:AGGREGATE [MERGE FINALIZE]                             |
| |  group by: bool_col                                     |
| |  having: bool_col = FALSE                               |
| |                                                         |
| 02:EXCHANGE [PARTITION=HASH(bool_col)]                    |
| |                                                         |
| 01:AGGREGATE                                              |
| |  group by: bool_col                                     |
| |                                                         |
| 00:SCAN HDFS [functional.alltypes]                        |
|    partitions=24/24 size=478.45KB                         | <-- Predicate could be attached to the scan
+-----------------------------------------------------------+
Returned 15 row(s) in 0.04s