"Volker Hetzer" <volker.hetzer_at_ieee.org> wrote in message news:<bm41va$278$1_at_news.fujitsu-siemens.com>...
> "Thomas Kyte" <thomas.kyte_at_oracle.com> schrieb im Newsbeitrag news:7b0834a8.0310090722.7512c04_at_posting.google.com...
> > Richard Kuhler <noone_at_nowhere.com> wrote in message news:<ROXgb.2708$Z86.2251_at_twister.socal.rr.com>...
> > > I'm reading "Effective Oracle by Design" by Thomas Kyte. He seems to be
> > > advocating using GATHER_SYSTEM_STATS _instead_of_ tuning
> > > OPTIMIZER_INDEX_CACHING and OPTIMIZER_INDEX_COST_ADJ directly. I can't
> > > find anything in the Oracle documentation that suggests that the
> > > presence of system statistics causes the optimizer to ignore the other
> > > settings. While I intend to do some experimentation, I'm wondering if
> > > anyone can make a definitive statement in this regard. Or, am I just
> > > misinterpreting the passage?
> > >
> > > Anyone have an opinion or real world experience to share on this?
> >
> > well -- 'instead of tuning' doesn't mean 'oracle will ignore these'.
> >
> > What I was saying was that you might well discover you don't need to
> > tweak them, if you have good system statistics.
> So, what effect do they have in the presence of statistics?
that is when they work "best"
without valid object statistics -- the CBO doesn't stand a chance.
>
> Lots of Greetings!
> Volker
Received on Thu Oct 09 2003 - 18:21:20 CDT
