On 25.1.2015 02:04, AlexK987 wrote:
> Tomas,
>
> Thank you for a very useful reply. Right now I do not have a case of
> poor performance caused by strong data skew which is not properly
> reflected in statistics. I was being defensive, trying to prevent
> every possible thing that might go wrong.
OK. My recommendation is not to mess with default_statistics unless you
actually have to (e.g. increasing the value on all tables, withouth a
query where the current value causes trouble). It increases time to plan
the queries, collect statistics (ANALYZE / autovacuum) etc.
regards
--
Tomas Vondra http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services