Laurie Harris

Background

About Laurie Harris

Building a Data Warehouse The Kimball and Inmon approaches both have their advantages. Kimball is more presentation-oriented, with a terrific iterative approach and the efficiency of conforming dimensions. Inmon's method pays more attention to data acquisition and storing data efficiently. My choice: a hybrid approach. We need the iterative development cycles so we can deliver project results more quickly, using an Agile methodology. But we also need a core data store that is fully normalized and maximally documented so that we truly understand the data. Why not build the core data warehouse in 3rd normal form, iteratively? Build what we need for a few urgently-needed data marts, then build the marts or cubes themselves. Deliver! Then reap the rewards - confidence from your users that you actually CAN deliver, feedback to enhance the next iteration, and some conformed dimensions that can be reused in the next cycle.