Re: Q: NUMBER OF DUPLICATE IN

In article <3upi5h$ald@cssun.mathcs.emory.edu> timothy.hood@sltrib.com "Timothy Hood" writes: <snip> > 65536 (or maybe 65535). However, keep in mind that indices with highly > repeated values tend to become poor performers. 'Tend' is a massive understatement. My experience (based on confessidly ancient hardware) is that major problems can occur long before the above theoretical value (for repeated values in indicies) is reached, and is particularly noticable in insert, update & delete. I've also found that 'bcheck' thinks the index is broken long before the above number of duplicates is reached, and seems to repair it but doesn't. Therefore my advice (from University of Life) is: NEVER create an index with massive numbers of duplicates. If an index is REALLY required with a column with lots of duplicates as the initial value (maybe because you need to find the few different values very quickly, or because it's a status and will move through a progression to a final value which is nearly alwasy the same), append an additional column which is unique or fairly unique thorughout the table. -- ============================================================================ Sally Woolrich | This mail contains my personal sally@excelsis.demon.co.uk | views not those of my employer! ============================================================================