[IGNITE-15054] Use statistics for joins

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: sql
• Ignite Flags: Docs Required, Release Notes Required
• Assignee: Alexander Belyak
• Reporter: Alexander Belyak
• Created: 05/Jul/21 10:28
• Updated: 27/Dec/21 15:01

Description

Optimize join selectivity estimation using collected underlying tables statistics. Mostly, it's about equals conditions on join: for common cases when tbl1.col1 = tbl2.col2 presented and statistics available for both columns - optimizer can use additional conditions like (col1.maxValue >= col2.minValue and col1.minValue <= col2.maxValue).

For example: col1 contains numbers from 1 to 10 while col2 contains numbers from 1 to 100. So we can expect that only 10% of rows in tbl2 were selected.

It's all just a hypothetical option to dig in.

Issue Links

• is blocked by: IGNITE-14307 Calcite. Use statistics in cost model.