On Wed, Jul 25, 2018 at 11:14 AM, Thomas Güttler <guettl...@thomas-guettler.de> wrote: > > AFAIK powa is based on pg_stat_statements not on statistical samples. > But maye I am wrong.
Indeed, it's based on pg_stat_statements, but other extensions are supported too. Since pg_stat_statements already provides cumulated counters, there's no need to do sampling. But if you're interested in wait events information for instance, it supports (in development version) pg_wait_sampling extension, which does sampling to provide efficient and informative informations.
