> > Did you pay attention to the estimated number of groups (ie, estimated > output rowcount for the aggregation plan node) while fooling around with > the statistics? How does it compare to reality, and to 9.5's estimate? >
I'm re-doing the tests and paying attention to that now. With statistics = 100, the under / over estimations change only slightly. Nothing that drastically alters anything: https://explain.depesz.com/s/GEWy With statistics = 1000 pretty much the same as above: https://explain.depesz.com/s/6CWM So between 9.5.5, 9.6.1, none of the stats changed in a noticeable way. Changing the statistics target on 9.6.1 slightly altered the estimates, but nothing to write home about. All have some significant deviations from actual row counts in the part of the query which is making the query slow.
