Equally Easy in Either Approach
Making the Final Choice
So how do senior management and the enterprise BI system designers resolve the final deployment conundrum: Dimensional relational or OLAP? As I hope you appreciate, there is no slam dunk answer since there are significant advantages and disadvantages for both approaches. But let’s consider two extremes. If you are a large distributed enterprise with a number of different database vendors and you are struggling to establish more commonality across your BI deployments, as well as creating an enterprise wide pool of BI development expertise without being beholden to any single vendor, then I recommend dimensional relational. On the other hand, if you are looking for the most potent single-subject solution with high performance and killer analytics, and you are confident that you can source the development expertise you need, then I recommend OLAP supported by underlying dimensional relational schemas. Otherwise, it depends.
Ralph Kimball, founder of the Kimball Group, teaches dimensional data warehouse and ETL design through Kimball University and reviews large warehouses. He has four best-selling data warehousing books in print, including The Data Warehouse ETL Toolkit. Write to him at [email protected].