Re: Full table scan used instead of index?

From: Thomas Hyldgaard <hyldgaard_at_hotmail.com>
Date: 15 Aug 2001 00:01:42 -0700
Message-ID: <62b0a667.0108142301.79f26967_at_posting.google.com>


Hi Jim,

Yes, I've tried analyze table test compute statistics; It helps immediately after the analyze, but after 15-30 minutes then the performance problem appears again. And still the execution plan says it uses full table scans.

I also experimented with the index by using create index test_index on test(c)

   tablespace TOOLS /* tablespace should be modified */    storage (initial 12280 K next 3080 K );

but so far without any luck.

I'm wondering if this problem is related to the used column data type or more generally to the way the index is created?

Best regards
Thomas

"Jim Kennedy" <kennedy-family_at_home.com> wrote in message news:<gele7.429580$p33.8467733_at_news1.sttls1.wa.home.com>...
> Did you analyze the table?
> analyze table test compute statistics;
> analyze index IndexName compute statistics;
> Jim
> "Thomas Hyldgaard" <hyldgaard_at_hotmail.com> wrote in message
> news:62b0a667.0108141228.5af50590_at_posting.google.com...
> > Hi, I have a table with approx. 419241 records. I'm using Oracle
> > 8.1.6.0 on Windows 2000.
> >
> > A query like:
> >
> > select * from test where c='0123456789012345'
> >
> > takes extremely long time (18 sec), and in the execution plan it says
> > that a full table scan is used. The strange thing is that it does not
> > use the index which is defined for the column - even if a hint is
> > used!
> >
> > The table is creates as:
> > CREATE TABLE(c raw(32),t raw(32));
> >
> > Can indexes work on the raw(32) data type?
> >
> > Best regards
> > Thomas
Received on Wed Aug 15 2001 - 09:01:42 CEST

Original text of this message