Hi Gail,

(I promise I'll give up soon) - Am I missing something (as some colleagues have already suggested).

If an identity column is used for the clustered index (CI) and NC indexes use the identity column (key) rather than a RID as the retrieval pointer then doesn't that mean that when the NC index is searched and the key (Identity column) is retrieved, it then needs to traverse the CI tree down to the leaf level to retrieve the actual row required as the 'identity column' alone wouldn't physically locate the row, it's just part of the data (leaf level of the CI)?

Jim