fragment table by date?

Hi! My statistics table is filling and I am looking for ways to improve search performance. I get about 25000 records (about 5MB) per night loaded into my statistics table. During the day the data is accessed by one user only (so far), making queries almost exclusively within one month's dates. If I keep one year's data (or more) does it make sense to fragment the table by expression on the date (month) to eliminate always 11/12 of the data? I have three disks (+ three mirror disks) available and thought of creating 12 dbspaces in a round robin scheme across the disks. What are the pitfalls to be considered? One of my problems is that I can't fragment by MONTH(my_date) so the fragmentation expression becomes an ugly list of " senddt <='01/31/1998' and senddt >='01/01/1998'" statements. Any suggestions? Thanks for your interest - and help. Cheers. Ralph. "If you rush a Miracle Man you get rotten miracles." ---------------------------------------------------- -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading