IBM Cognos User Manual page 21

If you have multiple data sources, you must relate the dimensions and levels of
your model to the data source that contains the columns to be used for each
measure.
Your model is more effective if your measures are applicable to more than one
dimension. For example, if your dimensions are products, locations, and
customers, your measures should bridge these dimensions.
2. Specify a Time Dimension
To ensure that your users can make period-to-period comparisons and visualize
trends over time, choose a time dimension that reflects and synchronizes
accounting periods and reporting schedules.
In most cases, your requirements are met by models based on the calendar or
fiscal year. Month, Quarter, and Year categories can be supplemented by
relative time categories automatically generated by Cognos Transformer, such
as YTD Growth, the percent-growth year-over-year.
If your organization uses particular time periods, such as lunar weeks and
months, or three 8-hour shifts per day, Cognos Transformer supports the
definition of custom time dimensions. Even if your query objects originate in
Framework Manager, you should import the necessary time-related items into
Cognos Transformer, and then define your time dimensions there.
3. Select the Data to be Modeled
You begin by identifying the data sources that contain the data for the model
you want to create.
Suppose that information about your customers is stored in a Customers table
and information about your products is stored in a Products table. Related
tables called Customer_Details and Product_Details provide additional
information about customers and products. Order information is stored in two
tables called Orders and Order_Details.
In keeping with good design practice, you decide to set up the Customers,
Customer_Details, Product, and Product_Details tables as structural data
sources, to provide the information that Cognos Transformer uses to build the
Customers and Products dimensions in your model.
The information about transactions is stored in the Orders and Order_Details
tables. For efficiency, you decide to combine the information in these tables into
a single data source called Order_Info.
The Order_Info data source contains the following information, all of which
you use to associate sales with particular customers and products:
v The order dates generate categories for the time dimension.
v Data about customers and sales representatives generates the header
information.
v The product, order quantity, and sales amount for each line item in an order
provide the sales measures.
v The cost of the order and discounts applied to it provide supplementary fact
data.
Example - Your Prototype Sales Model, on Paper
You can create an initial dimension map on paper, to make sure you have
identified all of the dimensions, levels, and categories needed in your PowerCube.
The measures to be associated with this dimensional hierarchy are Sales, Order
Qty, Cost, and Discount.
You map the dimensions of your prototype as follows:
Chapter 2. Planning Your Model
7