I've been doing quite a bit of work with point-only lat/lon data on large tables, and I'm now reached the question of...
Why did Oracle use an R-Tree (MBR) index format - on something that has no MBR (since a point by definition has no area)?
I have more comments (based on work I've now done using dual-column B-Tree indexes instead) - but I'm just curious from anyone with an Oracle POV before I go into that to help others that may be struggling with similar index build performance issues.
Bryan