Hope you may be able to help.

I am developing a fairly standard star schema with a central fact
table and 5 dimension tables. All dimension tables contain a surrogate
primary key. It is a composite of these keys that I plan to create the
fact table primary key with.

Am I correct in saying that if i use the date key as the first in the
composite and create a clustered index on this primary key i will have
data sorted by date always. My concern is that a a fact could contain
a small customer no and product no therefore would the composite key
be sorted on date or not just a combination of these values.

Alternatively what if i cretated the composite key as unique only and
then carried out a clustered index on date key, customee key etc? Is
there a max no of columns I shouldn't exceed for a clustered index?

Hope this makes sense.

Any help is most appreciated.

Cheers

Daniel

"Daniel Stone" <danielstone75 (AT) hotmail (DOT) com> wrote

Why not design your fact table with a surrogate key that will be your
primary key?