Applying Agile to Data Warehouse Design
I’m in the process of reading a book on Agile database warehouse design titled, appropriately enough, Agile Data Warehouse Design and by Lawrence Corr.
While Agile methodologies have been around for some time – going on two decades – they haven’t permeated all aspects of software design and development at the same pace. It’s only in recent years that Agile has been applied to data warehouse design in any significant way.
I’m sure many Agile consultants have worked on projects in the past where they were asked to come up with a complete design up-front. That’s true with data warehouse projects too where a client’s database team wanted the entire schema designed up-front – even before the requirements for the reports the data warehouse would be supporting were identified. What would appear to be driving the design was not the business and their report priorities, but the database team and their desire to have a complete data model.
While Agile Data Warehouse Design introduces some new methods, it emphasizes a common-sense approach that is present in all Agile methodologies. In this case, build the data warehouse or data mart one piece at a time. Instead of thinking of the data warehouse as one big star schema, think of it as a collection of smaller star schemas – each one consisting of a fact table and its supporting dimension tables.
The book covers the basics of data warehouse design including an overview of fact tables, dimension tables, how to model each and as mentioned, star schemas. The book stresses the 7-Ws when designing a data warehouse – who, what, where, when, why, how and how many. These are the questions to ask when talking to business to come up with an appropriate design. “How many” is applicable for the fact tables, while the other questions apply to dimension table design.
Agile Data Warehouse Design stresses collaboration with the business stakeholders, keeping them fully engaged so that they feel like they are not just users, but owners of the data. Agile Data Warehouse Design focuses on modeling the business processes that the business owners want to measure, not the reports to be produced or the data to be collected.
I still have a way to go before I’ve finished the book and then applied what I’ve learned, but so far, it’s been a worthwhile learning experience.