By user741803 on Mar 24, 2011
A past marketing tag line for Oracle BI EE was, “An information modeling approach.” In my experience, thinking in terms of information modeling rather than just SQL is one of the keys to developing the Common Enterprise Information Model (CEIM) successfully. When I’m designing a query factory, I need to think in terms of the models provided in the data sources, the simplified model I want to present to the users, and the schema shape transformations from one to the other. The data shape changes over and over throughout the query cycle, from the source schema, to the physical result sets, to the logical table sources, to the business model, to the Logical SQL (LSQL) result set, to the visualization. This is sort of like a*(b + c) = a*b + a*c in algebra, where the different shapes are good for different purposes, but give the same final result. Denormalized, normalized, star, snowflake, multicube, hypercube – the same information can be presented in all of these shapes.
The purpose of this post is to help you become familiar with dimensional shape-changing so you can understand the modeling patterns in the following posts. This will be important for understanding how to design the mappings in the CEIM – being able to recognize the dimensional structures within a larger physical schema, and then map them to the dimensional business model. There is also a bit of review of dimensional concepts and terminology.[Read More]