Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Query tuning experts: Any hope for tuning query with 5+ joins and a sort?
Tue, 5 Nov 2002 13:03:23 +1000, Richard Foote said (and I quote):
>
> Unique/Non Unique, it makes no difference in that an index range scan
> reading heaps of rows is an index range scan reading heaps of rows. To be
> avoided *if* it costs more than other options.
Good point.
> >
> > Impressive. What I don't get is why the stats for the 10% full scan are
> > not returning the same number of blocks visited. 10% or not, it should
> > have been the same full scan.
>
> Because the second query performs a much smaller disk sort and hence
> accesses a reduced number of temp blocks.
Yup. Makes sense.
>
> Go for it. I look forward to hearing what conclusions you come to.
>
Gimme a coupla days. In the middle of doing some design work for the RAAF and I'm floating around between North Sydney and the Blue Mountains... Back at my desk soon (I hope). BTW, thanks for the feedback and the numbers.
-- Cheers Nuno Souto nsouto_at_optusnet.com.au.nospamReceived on Tue Nov 05 2002 - 05:14:36 CST