posted 3/8/2010 by MikeMollenhour - Views: [1726]
Probably the most no-brainer in a DataWarehouse scenario is to index the heck out of the Date dimension. The cost of an index involves a couple of things:
1. How many inserts are performed on the table?
2. Fragmentation that may occur.
3. How much gain can be achieved by selects against the tables.
So first off date dimensions only need a row per day so very little inserts occur. Fragmentation can be maintained by disabling and rebuilding the index. Last but not least the date dimension is arguably one of the most hit tables in a data-warehouse. So as for me I will continue to index the date dimension as much as possible.