Neil Conway <neilc@samurai.com> writes:
> Tom Lane wrote:
>> This is something we need to consider, but we'll need more evidence
>> before making a choice. One thing that we have very little data about
>> is how much difference it makes in the quality of planner choices.
> Right, but is there a practical way to actually get this data?
I haven't thought of one yet, but perhaps someone will have an idea.
>> Also, I would expect that larger stats targets would slow down the parts
>> of the planner that look at the stats, since there are more data values
>> to examine. I do not have any numbers about this cost though --- do you
>> want to try to get some?
> Given the magnitude of the change (25 data elements versus 10), I
> wouldn't expect this to produce a major change in the total runtime of
> the optimizer.
I wouldn't either, but if we need to raise the stats target to 100 or
1000 to make a meaningful difference, then the question becomes more
urgent.
regards, tom lane