Doug Chamberlin wrote:

> Before you go too far with this discussion, would someone define what
> clustered indexes are and what problem they were created to solve?

I thought I just defined them - either putting the whole of the record
at the leaf level of an index, or clustering records by key value.

The problem that clustering is supposed to ...solve is too strong a word
... ameliorate, I guess, is the cost of reading an index entry, bouncing
over to the a data page, reading the page, returning to the index,
reading the next entry, returning to a different data page, reading the
record and so on, with the likely effect that each data page is read
several times. Which problem our intermediate bit maps avoid, completely.


Regards,


Ann