Now that we have listed the data attributes that might become dimensions in pre-summarized tables, it becomes apparent that pre-calculating all of the possible dimensions would involve many possible combinations. But how many summary tables? Let’s start with a simple example. Assume that we have four attributes: A, B, C, and D. We would have six possible combinations of attributes, namely, A-B, A-C, A-D, B-C, B-D, and C-D. As it turns out, the following formula can be used to determine the number of possible tables:

Number of dimension summary tables = (n)! / (n-2)! * 2

Where n = the number of dimension attributes

For four dimensions, we can quickly compute that there are six summary tables that can be built against the database, as follows:

Number of dimension summary tables = (4)! / (4-2)! * 2

Number of dimension summary tables = 24 / 4

Number of dimension summary tables = 6

However, in our example, we have 10 dimensions. We can compute the number of possible combinations of these attribute tables, as follows:

Number of dimension summary tables = (10)! / (10-2)! * 2

Number of dimension summary tables = 3,628,800 / 40,320 * 2

Number of dimension summary tables = 3,628,800 / 80,640

Number of dimension summary tables = 45

Note: The number of combinations equals 45, but the real number of tables we need is 90 since we must aggregate two facts. This is because we have to perform the calculations twice--once to summarize the total_cost fact and again for the quantity_sold fact.

Also, we will not be able to create summary hierarchies like our Excel pivot table did in Chapter 5, so each aggregate table will be isolated and referenced by some permutation of the dimension names. If we are going to create 45 tables for each fact, we need to provide meaningful names to our tables. Sample table names might include MONTH_BY_YEAR, MONTH_BY_QUARTER, and MONTH_BY_CUSTOMER_NAME.