Another issue that is milder but chafes me is the concept that the PK field is usually the clustered index. Certainly this is default, and certainly this sometimes makes sense, but in large tables it may well be that other fields make more sense. If one table joins to another table frequently and that join returns multiple rows, like a claim table joined to a notes table on claim ID, it may make sense to cluster the notes table on claim ID since that will allow the query that paints the notes screen on the claim system to work far faster (and since note ID is almost never used for any queries). Again, one needs to think about what makes sense performance-wise and not just blindly make all primary keys clustered.