Hi,
Points taken - "hard to believe" (whole clustered index value in non clustered index) - I'm surprised that non clustered indexes don't also use the RID to provide a more efficient use of space.
I know space isn't so much of a problem nowadays however disk reads/writes are still (generally) the most time consuming function that a PC etc performs, so reducing the physical data storage size can overall, increase the data density and reduce the amount of I/O operations required, particularly head movement.
This is of course countered by the need to update all non clustered indexes when a row is moved to a different page as the result of a split operation but this (as you stated in your article) should be minimised by the clustered key design (i.e. stable, non changing).
"Article suggest a full able scan required".. my sincere apologies, I have re-read the para and you didn't make that suggestion - I mis-read the content of the SCAN method of accessing a table....
Sorry if I misled anybody or gave the impression that I thought the article was inaccurate, this wasn't my intent.
I will be reading part 3 when I get the opportunity....
Jim
Trainee Novice:w00t: