On Tue, Sep 30, 2014 at 8:34 AM, Graeme B. Bell <grb@skogoglandskap.no> wrote:
>
>>> The existing cost estimation
>>> code effectively assumes that they're perfectly uniformly distributed;
>>> which is a good average-case assumption but can be horribly wrong in
>>> the worst case.
>
>
> Sorry, just an outsider jumping in with a quick comment.
>
> Every year or two the core count goes up. Can/should/does postgres ever attempt two strategies in parallel, in cases
wherestrategy A is generally good but strategy B prevents bad worst case behaviour? Kind of like a Schrödinger's Cat
approachto scheduling.
> What problems would it raise?
Interleaved I/O, that would kill performance for both plans if it
happens on rotating media.