At 7/21/2001 03:10 AM (Saturday), Joseph wrote:

>I would just like to request for comments on my article below so that I
>will know if my views are correct or not.

Excellent description of the problem your investigation into it!

I know you are much more interested in Jim's comments but here are mine.

The only problem I have with your article is in the use of the term
"selectivity" as in "avoid selectivity". Selectivity is a measure of an
index's ability to select a record. If there are only 3 index values spread
over millions of records then the index's ability to select a specific
record using that record's index value is poor. If the index is unique then
its selectivity is excellent.

So, I would rephrase your article to say "avoid poor selectivity" and
attempt "good or high selectivity". You see ALL indexes have SOME
selectivity measure.

A small but important clarification.