[IMPALA-10185] Leverage boolean statistics for computing selectivity

• Type: Improvement
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Affects Version/s: Impala 3.4.0
• Component/s: Frontend
• Epic Color: ghx-label-7
• Assignee: Shant Hovsepian
• Reporter: Shant Hovsepian
• Created: 23/Sep/20 04:47
• Updated: 21/Dec/20 19:52

Description

As part of computing stats for a boolean column we record the value counts.

numTrues, numFalses, numNulls.

However when calculating cardinality for boolean column predicates the selectivity of 1/3 is always used instead of factoring the value distribution recorded in the stats.

Issue Links

• is related to: IMPALA-7601 Improve cardinality and selectivity estimates