Re: Wrong number of rows in tkprof?

"Rene Nyffenegger" <rene.nyffenegger_at_gmx.ch> wrote in message news:Xns90FC17041F68gnuegischgnueg_at_130.133.1.4...
>
> >The number you are seeing is the cardinality as computed by the
>
> If you say 'computed', do you mean, the optimizer sort of predicts
> a result, regardless of the actual value?
>
> >optimizer. How did you analyse those two tables, if at all?
>
> I didn't analyze the tables at all. That would have been the next
> step I wanted to do.
>
>
> >Hth,
> >
> >Sybrand Bakker, Senior Oracle DBA
>
> Rene
>
> --
> Recherchen im Schweizerischen Handelsamtsblatt:
> http://www.adp-gmbh.ch/SwissCompanies/Search.php3

Yes, at least it does in 7.3.4. I will check soon whether it still applies to 8.1.7
The number of rows in tkprof usually agrees with the computed cardinality results you get from
alter session set events='10053 trace name context forever', which will basically show you (or try to show you) why the optimizer made those stupid decisions.
Needless to say: those computed values are quite often *wrong*. If you didn't analyze the affected table it is not surprising they are wrong: the values are estimated in that case. But I did analyze my tables (computed)

Hth,

Sybrand Bakker, Senior Oracle DBA Received on Tue Aug 14 2001 - 14:44:06 CDT