On Tue, 2007-02-06 at 12:10 +0900, ITAGAKI Takahiro wrote:
> "Simon Riggs" <simon@2ndquadrant.com> wrote:
>
> > > Actually, given what we've just learned --- namely that choosing these
> > > values at random is a bad idea --- I'd want to see a whole lot of
> > > positive evidence before adding such a configuration knob.
> >
> > 3. assemble performance evidence
> >
> > Step 3 is always there for performance work, so even if you don't
> > mention it, I'll assume everybody wants to see that as soon as possible
> > before we progress.
>
> There was a performance evidence using TOAST in order to partial updates.
> It added a flag of force toasting. The toast threshold suggested now is
> more flexible than it, but I think it is one of the evidences.
>
> Vertical Partitioning with TOAST
> http://archives.postgresql.org/pgsql-hackers/2005-12/msg00013.php
>
Apologies to Junji-san. I'd thought my idea was original, but it seems
we think along similar lines.
That is the kind of performance gain I see possible.
-- Simon Riggs EnterpriseDB http://www.enterprisedb.com