>Larry,
>
>One important thing you're forgetting about here is that non-clustered indexes store a pointer to the location that record can be found. If the table has a clustered index, this is used as the pointer. If the table is a heap, it uses an RID to locate the record. (File#:Page#:Slot#)
>

Hi Travis,
So you are saying that a pointer to the index node is faster than a pointer to the physical location of the row in the table? I'll but that.

I may have to go back and check performance on tables where I didn't specify one. If I recall, the numbers weren't bad so I didn't worry about tweaking the performance.