You can improve SQL server performance if you avoid creating tables in a monolithic fashion. By monolithic I mean tables that have dozens and dozens of fields that are not often used together. Some shops lump all data associated with a key field (or fields) into one table. A better structure is to make multiple tables, each containing only the relevant data that is used together, with each table having the same key field. For example, instead of using one “Big Table,” use three smaller ones:

 BIG_TABLE   --becomes-- PART_TABLE	MTD_TABLE	YTD_TABLEpart_num (key)		part_num (key)	part_num (key)	part_num (key)	-------			-------		-------		-------part_desc		part_desc	mtd_ship	ytd_shippart_wt			part_wt		mtd_sales	ytd_salespart_uom		part_uom	mtd_return	ytd_return...mtd_shipmtd_salesmtd_return...ytd_shipytd_salesytd_return