Re: benchmarking the query planner

On Fri, 2008-12-12 at 02:23 +0000, Greg Stark wrote:

> The existing sampling mechanism is tied to solid statistics. It
> provides the correct sample size to get a consistent confidence range
> for range queries. This is the same mathematics which governs election
> polling and other surveys. The sample size you need to get +/- 5% 19
> times out of 20 increases as the population increases, but not by very
> much.

Sounds great, but its not true. The sample size is not linked to data
volume, so how can it possibly give a consistent confidence range?

-- Simon Riggs           www.2ndQuadrant.comPostgreSQL Training, Services and Support