[postgis-users] Test bench results on peformance/limits of postgis/postgresql?

[Thomas Lefort]

Hello,

I am wondering if there is any performance test bench results on the use of
postgresql/postgis for geometry intersections.

I have this set up: several millions of polygons, with up to 20-30 points
each (satellite instrument footprints) with a start and stop time each, and
I need to make a query to retrieve all of those polygons that intersect an
area of interest and withing a given time range (and possibly other type of
filters such as name of sensors etc...). I have a working configuration
which is rather fast wrt to the actual request made (number of matching
results) but there is not that many footprints, eg several 100000. I
created a GIST index on my geometry and also indexes on various other such
as sensor name.

My question is how will this scale? I could randomly add many footprints
and make some tests, but if there is an interesting study on this, in
particular recommendations on database configuration, indexes, hardware,
etc... I would be very interested to read them.

Thanks,

Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20120710/bc3c4c71/attachment.html>