Re: SQL vs. Oracle

If you have skewed data (e.g. 90% of your data has one value, the other 10% has 5000 values), using an index definitely can be bad if you need to access the single value that's in the 90% range.

--
Terry Dykstra
Canadian Forest Oil Ltd.
Check out Sybase Developer's Network: http://www.sybase.com/sdn

Daniel <delj_at_flash.net> wrote in message news:86qio2$re_at_journal.concentric.net...
> Don't know 'bout you but most of my tables are very large. If it isn't
> using an index, it is doing a full table scan and beating my hard drives
to
> death. In addition to beating my hard drives, it takes much longer to do
a
> full table scan. Cost based optimization would try to use indexes, not
> rules based. First rows is good if you want to use hints. See my
original
> example. The RDBMS should have used the index for an inner join. The
index
> was there. It is faster and more cost efficient to use the index, why
> should I have to tell the most powerful and expensive RDBMS on the planet,
> to use an index that is available for the field referenced in my where
> clause. The only time using an index is bad, is when the index is
corrupt.
> Can you think of an example to support your statement that sometimes it is
> bad?
>
> Daniel
> Austin, TX
>
> > What makes using an index so good ?
> >
> > Sometimes using an index is good - sometimes its bad.
> >
> > If you want to use the index in almost all situations, make your
> > optimiser rule based or first rows based.
> >
> > HTH
> > --
> > ===========================================
> > Connor McDonald
> > "These views mine, no-one elses etc etc"
> > connor_mcdonald_at_yahoo.com
> >
> > "Some days you're the pigeon, and some days you're the statue."
>
>
Received on Thu Jan 27 2000 - 17:22:04 CST