> mark.kirkwood@catalyst.net.nz wrote on 03.05.2013 00:19:
>> I think the idea of telling postgres that we are doing a load is
>> probably
>> the wrong way to go about this. We have a framework that tries to
>> automatically figure out the best plans...I think some more thought
>> about
>> how to make that understand some of the more subtle triggers for a
>> time-to-do-new-plans moment is the way to go. I understand this is
>> probably hard - and may imply some radical surgery to how the stats
>> collector and planner interact.
>
> I wonder if "freezing" (analyze, then disable autovacuum) the statistics
> for the large number of rows would work.
>
>
>
I'm thinking that the issue is actually the opposite - it is that a new
plan is needed because the new (uncomitted) rows are changing the data
distribution. So we want more plan instability rather than plan stability
:-)
Cheers
Mark