Re: STATE columns - bitmap index or not?

From: Howard J. Rogers <howardjr_at_www.com>
Date: Tue, 13 Feb 2001 19:53:18 +1100
Message-ID: <3a88f5ff$1@news.iprimus.com.au>

If only I could get your book....

Bugger!
HJR "Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote in message news:982049805.15759.0.nnrp-10.9e984b29_at_news.demon.co.uk...
>
> Actually, there is a worked example in my
> book which demonstrates that clustering and
> rows per index value is more important.
>
> The actual example has 2,800 rows and a
> cardinality of 560 (i.e. 5 rows per value).
>
> Of course, the point up rate of change is
> very important - bitmaps on tables
> subject to almost any change tend
> generate huge amounts of contention,
> and fragment very badly.
>
>
> --
> Jonathan Lewis
> Yet another Oracle-related web site: http://www.jlcomp.demon.co.uk
>
> Practical Oracle 8i: Building Efficient Databases
> Publishers: Addison-Wesley
>
> Reviews at: http://www.jlcomp.demon.co.uk/book_rev.html
>
>
>
> Howard J. Rogers wrote in message <3a88e233_at_news.iprimus.com.au>...
>
>
> >But that aside, I believe there was a post ages ago (either from Jonathan
> >Lewis or Steve Adams, but my brain may be failing me utterly, and it
could
> >have been someone completely different) that suggested a cardinality of
50
> >is not unreasonable (it seems just a trifle high to me -I always teach a
> >cardinality of 20 is where it starts to get dubious).
>
>
>
Received on Tue Feb 13 2001 - 02:53:18 CST