Re: Table design decision

"Tarby777" <nick_williamson_at_mentorg.com> wrote in message news:1136989006.928023.79670_at_g44g2000cwa.googlegroups.com...
> Hi all,
>
> I've inherited a monster table, and have to make a decision about
> how/whether to normalize it.
>
> It has 50 FK columns, each one with a FK constraint and a supporting
> index. The worst thing from a design perspective is that in each row,
> only exactly one of those FK columns will ever be populated in any
> given row. All the columns get populated, but never in the same row. So
> as things stand, every row contains 49 nulls, and there are 49 index
> writes that are of little use because the application will never
> "select * from this_table where FK column is null".
>

...

nulls are not indexed

++ mcs Received on Wed Jan 11 2006 - 09:37:47 CST