Re: clustering factor and uniqueness

<yoram.ayalon_at_structuredweb.com> wrote in message news:1143752412.171160.252210_at_e56g2000cwe.googlegroups.com...
> thanks for the support
>
> followup - I rebuilt the index as a bitmap, and also corrected the bug,
> now i access it thru its correct type. I only caught ther bug as I was
> already rebuilding the index as a bitmap...
>
> the access is really fast now, even though the CF is of course the
> same. this tells me Oracle understands that since it has almost as many
> distinct keys as a PK it can ignore the high CF
>
> this table is updated in batches, in the OLTP is only selected, thus a
> bitmap idnex is feasible.
>
> now I have to rebuilt the index as a b-tree and check the difference. I
> wil post the results in case anyone is still interested
>
>
>
>
> I don't think I will rebuilt the table as along as I get results as of
> now. At least I have learned a lot about indexes, CF and the all the
> other good stuff. thanks again everone
>

For bitmap indexes, the clustering_factor column is irrelevant - it simply echoes the num_rows column of user_indexes.

The data distribution pattern is assumed to be 80% packed, 20% scattered for the purposes of calculating the effective clustering of bitmapped data. (See Oracle Wait Interface by Richmond Shee et. al.)

-- 
Regards

Jonathan Lewis
http://www.oracle.com/technology/community/oracle_ace/ace1.html#lewis

The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html

Cost Based Oracle: Fundamentals
http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html