=?ISO-8859-2?Q?pasman_pasma=F1ski?= <pasman(dot)p(at)gmail(dot)com> writes:
> Thanks for reply.
> I tested random changes and query runs fastest after:
> set seq_page_cost = 0.1;
> set random_page_cost = 0.1;
> cpu_operator_cost = 0.01
As a general rule, "optimizing" those settings on the basis of testing a
single query is a great way to send your overall performance into the
tank --- especially since repeating a single query will be heavily
biased by cache effects. You need to look at a representative sample of
all your queries across all your data.
regards, tom lane