Greg Stark wrote:
> clustering...
> That will only help if you're often retrieving moderately large result sets by
> one particular index. If you normally only retrieve one record at a time or
> from lots of different indexes then it probably won't really make much
> difference.
It'll also help for columns whose values are related in some way.
For example, clustering a table of addresses based on "zip code"
will help lookups based on city or county or state (presumably
because all the disk pages for a given city will be grouped
together within the disk pages for the zip codes within the city).