Am Donnerstag, 31. August 2006 15:06 schrieb Jeroen T. Vermeulen:
> Is there any kind of pattern at all to this problem? Anything
> recognizable? A few typical pitfalls?
If data is not distributed evenly, then any old WHERE foo = $1 is prone to be
the wrong plan for half of the possible values of $1. The more data you have
and the more it changes, the worse this gets.
--
Peter Eisentraut
http://developer.postgresql.org/~petere/