Rexem Feedback Extension:
The idea of a clustered index for pkeys is not arbitrary. This is just the default to group pkey. And clustered means that the values will be physically placed next to each other on the Sql Server 8k page, thus assuming that if you select a single value using pkey, you will probably be interested in its neighbors. I do not think it is a good idea to do this for pkeys, as they are usually unique but arbitrary identifiers. It is better to cluster more useful data. One clustered index for each table.
In a nutshell: if you can filter your query in a clustered index column (which makes sense), all the better.
source share