Re: About indexes and clusters
Date: Fri, 7 Nov 2003 09:54:00 -0800
Message-ID: <p8Rqb.6$FN1.123_at_news.oracle.com>
"Juan Pardillos" <sicotom_at_eresmas.com> wrote in message
news:6278687.0311070834.131981e4_at_posting.google.com...
> Hi everybody,
>
> I have some doubts about these topics. I hope you can help me:
>
> a) I have read that some queries can be solved by just querying the
> indexes, without actually retrieving data. I don't see this point.
> Could anyone give me an example?
Because each dereferencing of a table row given row address in the index leaf results in one logical io, which may translate into one random disk access.
> b) Why it is possible to define an index over several attributes (and
> the file is ordered taking all of them into account, in the order
> these attributes are specified), but it is only possible to define a
> cluster over a single attribute?
IMO, clustering of the data is not such a good idea. Clusters are source of bugs (as just yet another feature, that database implementors should take care of), and contrary to popular belief don't improve the performance. Received on Fri Nov 07 2003 - 18:54:00 CET