108 Solving Operational Business Intelligence with InfoSphere Warehouse Advanced Edition
After the data is populated, this new structure is ready for use by the BI
applications, such as IBM Cognos. Although these BI applications can directly
access the data within the database, there is still a more efficient and effective
mechanism that can be introduced to perform this task. OLAP modelling
provides a method by which these BI applications can communicate with the
users in a dimensional language closer to the way humans think, while at the
same time providing a business abstraction layer between the data warehouse
and these BI applications.
Within InfoSphere Warehouse, OLAP modeling is defined as a cube model,
which can then be implemented as a cubing service. Although a star schema can
begin to provide an abstraction to a relational database, it lacks the depth and
richness of OLAP objects such as hierarchies, levels, and derived measures and
attributes. OLAP cubes also provide an enhanced ability to capture business
rules and logic, which is not as easy or intuitive within a relational database.
You might ask, why is this called a cube model?
Questions asked by business analysis such as “what is the sum of all sales” or
“how many stores are there” can easily be answered by a normal SQL query of a
relational database. When taken in context of your logical data model, these
types of questions only relate to one dimension within the model. Business
questions, however, are generally of the form that ask about something (a
dimension)
by something (another dimension), and might be even by something
again. This form of multidimensional analysis is best suited to a cube model,
where data can be analyzed from multiple perspectives.
Consider that you have to analyze the sales data for a number of products in a
range of countries that you have stores, for all periods of time. To help visualize
the data from these multiple perspectives, think of a cube with each of the axes
representing a dimension. For example, consider country, product, and time.
There is also an implicit fourth dimension measure, which is the sales amount.
Figure 4-15 on page 109 depicts this cube, which itself has then been divided
into a number of mini-cubes. Each mini-cube contains a value (or measure) for
that specific intersection of the country, product, and time. As an example, the
darker mini-cube (shown in blue) represents the sales for the first quarter of 2008