On Mon, Sep 29, 2014 at 7:12 PM, Gavin Flower <gavinflo...@archidevsys.co.nz > wrote:
> > Would it be feasible to get a competent statistician to advise what data to > collect, and to analyze it? Maybe it is possible to get a better estimate on > how much of a table needs to be scanned, based on some fairly simple > statistics. But unless research is done, it is probably impossible to > determine what statistics might be useful, and how effective a better > estimate could be. > > I have a nasty feeling that assuming a uniform distribution, may still > end up being the best we can do - but I maybe being unduly pessimistic!. > As a semi-competent statistician, my gut feeling is that our best bet would be not to rely on the competence of statisticians for too much, and instead try to give the executor the ability to abandon a fruitless path and pick a different plan instead. Of course this option is foreclosed once a tuple is returned to the client (unless the ctid is also cached, so we can make sure not to send it again on the new plan). I think that the exponential explosion of possibilities is going to be too great to analyze in any rigorous way. Cheers, Jeff
