[SEDONA-189] Prepare geometries in broadcast join

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.3.0
• Labels: pull-request-available
• Assignee: Unassigned
• Reporter: Tanel Kiis
• Created: 04/Nov/22 13:43
• Updated: 11/Nov/22 05:35
• Resolved: 11/Nov/22 05:35

Description

With complex polygons using prepared geometries can improve query performance by an order of magnitude.

A test, where I had 1M points and 5k polygons, a simple broadcast join and count with ST_Contains had a performance increase from 40s down to 10s (4x improvement).

points.join(broadcast(polygons), expr("ST_Contains(polygon, point)")).count()

If the relative number of points to polygons increases, then the speedup gets better. For

points.union(points).join(broadcast(polygons), expr("ST_Contains(polygon, point)")).count()

it is 6x (70s -> 12s).

Issue Links

• links to: GitHub Pull Request #708