[postgis-users] Trying to figure out how statistics for the planner work

[Markus Schaber]

Hi, Steve,

Steve Wormley wrote:

> I've tried setting the statistics on the column to 10, 100 and 1000 and
> re-analyzing each time and I always end up with a plan with it expecting
> only one row.

Increasing the statistics target was my only suggestion :-(

So I step back and give way for the GIST experts here...

> Any suggestions, ideas, other ways to make this better?

Maybe the PostgreSQL Performance List <pgsql-performance at postgresql.org>
has something to offer here.

One question that comes into my mind is whether your points are spread
evenly? It could be that the area you queried is extremely dense,
whereas the rest of the area is populated lighter, and the statistics
engine does not get the distribution right.

HTH,
Markus
-- 
Markus Schaber | Logical Tracking&Tracing International AG
Dipl. Inf.     | Software Development GIS

Fight against software patents in EU! www.ffii.org www.nosoftwarepatents.org