Hi,
today I was reasoning on the speed up I got some years ago
on a non spatial db application by using clustered tables.
Clustered tables have the property of being physically
laid out on the same order of an index, or in some database,
being stored within that index. This boosts up tremendously
the access speed, in my app it ranged from 2 to 10 times
faster depending on the query.
Now, Postgres has a poor man version of clustering, in that
tables won't be kept clustered when data do change,
yet if your data does not change much, it's possible
to cluster them along the spatial index and rumors are
that there is a nice speed gain. More information here:
http://www.bostongis.com/PrinterFriendly.aspx?content_name=postgis_tut03
http://postgis.refractions.net/pipermail/postgis-users/2005-March/007260.html
Chris, did we ever tried this on Sigma? Since our data
is basically all read only, it may be an interesting test.
Cheers
Andrea