Performance myths: Clustered vs. non-clustered indexes I show a "redundant" non-clustered index outperforming the clustered index with the same key.