On 12 March 2016 at 22:00, Peter J. Holzer <hjp-pgsql@hjp.at> wrote:
> I don't think most people's data is perfectly distributed. But as you
> say most data is probably within some deviation of being perfectly
> distributed and as long as that deviation isn't too big it doesn't
> matter.
Is that how what I wrote came across? I was trying to say exactly the
opposite: that if, instead of assuming perfect distribution, you
assume a certain deviation from that distribution, you will end up
with plans that would win with perfect distribution losing out to
plans that are almost as good on that perfect distribution but behave
better on data that is not perfectly distributed.
Geoff