[postgis-users] Completely Non-Scientific Performance Test (CNSPT)

[Andy Turk]

<CNSPT>
We've been using a 2 million row database on a regular basis. Queries which 
pick up a hundred or two rows are difficult to measure without writing some 
timing code. Our "server" is a measly 500Mhz Athlon with only 256M of memory 
(but a nice fast SCSI disk).

I'm not sure how long the index took to build, because I did it one night and 
didn't time it.

One thing that I believe helps the performance we're seeing is that the 
geometries in the index are all roughly the same size and there isn't a lot 
of overlap.
</CNSPT>