On Mon, 2008-02-25 at 12:05 -0800, Joshua D. Drake wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Mon, 25 Feb 2008 11:36:56 -0800
> Jeff Davis <pgsql@j-davis.com> wrote:
>
> >
> > If there is any significant I/O latency for a single backend, it seems
> > like a context switch could be a win for processor utilization. It
> > might not be a win overall, but at least potentially a win.
>
> Do we want a 20% potential win or an 80% potential win?
>
> I would personally rather keep it simple, hard core, and data shoving
> as possible without any issue with scheduling etc..
>
Just a thought. After it's actually implemented it won't be hard to see
if it's a win.
Regards,Jeff Davis