On Fri, Jul 13, 2012 at 3:22 PM, Eoghan Murray <eoghan@qatano.com> wrote:
> This is with `enable_material=off`, with `enable_material=on` it also
> doesn't go for the Merge Join, but the Materialize step pushes it up to over
> 7,000ms.
I think this one could stem from what Tom observed, that the rowcount
estimate is way off.
It usually helps a great deal if you can get PG to estimate correctly,
be it by doing analyze, incrementing statistics or simply formulating
your query in a way that's friendlier for PG's estimator.
