Snowflakes are one of nature's most fragile things, but just look at what they can do when they stick together.
—Vesta M. Kelly
In Chapters 1 and 2, we learned how business intelligence can be used to support effective decision making. In Chapter 3, we began the search for the data that serves as the source for this business intelligence. In most cases, the data we need can be found in our OLTP systems.
On close examination, we discovered that an OLTP system is not a good candidate for the direct source of business intelligence. OLTP systems are designed for processing business transactions and storing the detail from each of these transactions. They are not optimized for delivering the aggregated information typically required for business intelligence.
The shortcomings of OLTP systems as a source for business intelligence led us to the data mart. The data mart is a relational database structure specifically designed for storing large amounts of historical data. Data must be copied from the OLTP systems into the data mart.
At the end of Chapter 3, we were left with a concern about the data mart. Namely, how do we improve performance when the user wants to view aggregate information derived from a large number of detail records? For the answer to this, we look to online analytical processing.