Cross-Functional Analysis and Denormalization
Cross-functional analysis becomes increasingly difficult as the data warehouse becomes more and more denormalized.
Data warehousing provides a means to build complex interrelated models for cross-subject area analyses in ways no other system can. These models can move beyond the traditional financial measures to begin interrelating internal process measures and customer-oriented measures as well. More importantly, these more sophisticated analytical models can begin to push from results‑oriented or outcome‑oriented measures toward measures directly linked to organizational activities. The following figure shows an example of interrelated measures from a credit-card model.
The problem with denormalized data for this example is that measures cross many different denormalized schemas. Even though Teradata Database is optimized to handle them, star joins are cumbersome to process, and making joins across multiple models also makes them extremely complex. In this particular example, there could easily be four or more distinct star schema structures.