What Is a Cube?

Cubes are typically built from a subset of a data warehouse and organized in a multidimensional structure, defined by a set of measures (summarized business information such as product sales) which are indexed by dimensions (business variables, such as time or location, on which the business facts depend). The illustration shows a schematic representation of a cube that contains two measures. “Packages” represents the number of packages imported and “Total Freight” represents the total shipping costs of those packages in a given quarter. The values of the measures depend on three dimensions: the route by which the imports reach their destination (the “Route” dimension), the locations where the imports are produced (the “Source” dimension), and the calendar quarter (the “Time” dimension) in which the products shipped.

The granularity of the cube is determined by the unit value (or key attribute) along each of the dimensions. For example, along the Time dimension, the key attribute is calendar quarter. Each cell in the cube contains the values of the two measures for specific values of Route, Source, and Time. For example, the cell in the lower left shows that 3,562 packages were shipped from South America by air in the first quarter. The total shipping costs of packages received from South America, by air, in the first quarter was $56,749.