Q:

How would you characterize the difference between Bill Inmon's philosophy on data warehousing and Richard Kimball's? As a new student to the subject, should I focus more on one philosophy over the other? And finally, who is the real father of data warehousing? I believe they both make the claim.

A:

Les Babusinski’s Answer: I’m not sure how helpful such titles are, but (for what it’s worth) I think most people would agree that Bill Inmon is the “father of data warehousing.” He was, after all, the first to define and champion the concept and is widely credited with coining the term “data warehouse.” Ralph Kimball, for his part, can rightfully claim the title of “father of business intelligence.” He codified the star schema and snowflake data structures, defined the concepts of data marts, dimensional hierarchies, “base” and “aggregate” metrics, “drilling,” etc. and, in short, developed the science behind modern analytical reporting tools. Both men have made immeasurable contributions to the field. As to their philosophies, they differ primarily in scale. In Kimball’s world, the data mart is the data warehouse, while in Inmon’s world the data mart is only one layer in the data warehouse. Inmon’s philosophy is most often associated with enterprise-level data warehouse efforts, while Kimball’s philosophy is most often associated with departmental data marts. Most modern incarnations implement the data mart model rather than the enterprise data warehouse because of the time, cost and risk of failure associated with developing the latter. However, in recent years, the proliferation of “silo” data marts throughout many enterprises has brought its own problems such as duplication of effort and costs, inconsistent metrics and dimensional hierarchies, lack of historical reporting capabilities, etc. This has led to the development of a hybrid approach (called “federated data marts”) which attempts to meld the best points of both philosophies into a cohesive whole. So, to be successful in the field of data warehousing, you really must understand both points of view.

Joyce Bischoff’s Answer: This question has become a religious issue in the data warehouse community. In my opinion, no single individual can claim to be the only father of data warehousing. In the 1970s and 1980s many people built systems for decision support purposes, which were later called information warehouses by IBM and data warehouses by others. Bill Inmon has been a superb communicator since the early 1990s on designing, building, implementing and managing the data warehouse environment. Ralph Kimball specializes in star schemas and dimensional modeling. Both people are very important contributors to the field. If you are beginning to study data warehousing, I suggest that you start with Bill Inmon’s books, classes and Web site, www.billinmon.com, and follow up with Ralph Kimball’s books, classes and Web site, www.ralphkimball.com. A successful approach to data warehousing must address much more than the design process. Issues involve data warehouse architecture, organizational culture, changing roles and responsibilities, data quality, data integration, etc. Many other writers, such as Claudia Imhoff, Doug Hackney and others, have made significant contributions should also be considered. Although star schemas and dimensional modeling are important parts of the data warehouse, there are limitations to Ralph Kimball’s dimensional data modeling techniques. For more information, see the white paper entitled “The Problem with Dimensional Modeling” at www.billinmon.com.

Scott Howard’s Answer: Both are very original and influential in our industry. Mr. Inmon is effective communicating the value of data warehousing and how it’s best used to solve specific business problems. He’s also very strong on the overall architecture. Mr. Kimball is the originator of the star schema and a master of multidimensional modeling. In my opinion, both are very worth studding as long as you limit your focus of each to what I’ve just shared as their major strengths. Who is the father of data warehousing? Hold off ordering the DNA test, as I’m about to complicate things! From my prospective, it’s Barry Devlin. I’m sorry for confusing you because modest and soft-spoken Barry is not as bold as the others in proclaiming parenthood. However, I remember listening to Barry advocating a more analysis-oriented way of constructing intelligent decision support systems he called Information Warehouses, reading his architectural papers and following his lead back in the 1980s. He did not publish his full book on Data Warehousing until just after Misters Inmon and Kimball, perhaps weakening his potential claim of fatherhood, but those of us at this game for a long time, remember the pioneers. Now I’m not trivializing the great contributions of the other influential data warehousing pundits, I’m just pointing out a lesser-known innovator. Does it really matter who was first? You were correct in asking, on whom should I focus. A great BI system will be a composite result of the best teaching and practices, not focusing on just one. This opinion is considered heresy by those who worship at the altars of a single BI prophet, but I find some distinct value in all of them. In addition to Inmon and Kimball, I strongly recommend studying Devlin’s teachings. Check out his Data Warehouse, From Architecture to Implementation, ISBN 0-201-96425-2; it’s easy to follow and innovative. Barry’s still innovating as he’s now advocating completing the BI cycle with feedback into the DW, measuring the effectiveness of both the decision-making process, including the business intelligence system and that of the decision- maker. This is a very logical progression, but also a very sensitive one in that decision-makers don’t like to be measured. I’d love to have the foresight to predict how this will end. If you are a project lead or a BI project manager, I strongly recommend Data Warehouse Project Management by Sid Adelman and Larissa Moss, ISBN 0-201-61635- 1. It provides great practical project management guidance in addition to a technology primer for those new to the field.

Chuck Kelley’s Answer: I don’t think that Ralph (not Richard – he was “The Fugitive”) Kimball has every made the claim to be the father of the data warehouse. Bill is recognized as the “father of data warehousing.” I think conceptually there is very little difference between Bill and Ralph’s data warehouse. The big difference is how and where the data is stored. Bill believes more in the large monolithic data warehouse (or distributed data warehouse, which is different than data marts) and Ralph believes in the sum of the data marts is the data warehouse as long as there are conformed dimensions. Conformed dimensions, in my opinion, are what make the sum of the data marts the equivalent of having a quasi-distributed data warehouse. Hence, they are basically the same in philosophy, which is all you can ask for! Everyone will choose a different way to implement.

Larissa Moss’s Answer: Bill Inmon's philosophy embraces any and all types of decision support solutions, whether they are operational, tactical or strategic as well as ad hoc queries, canned detailed reports and multidimensional trend analysis reports. Bill Inmon supports both the relational (two-dimensional) database design as well as the star schema (multidimensional) database design – whichever is the more appropriate for the specific decision support requirements. Ralph Kimball on the other hand concentrates on the multidimensional aspects of database design and reporting. In that sense the term "data warehousing" has a more limited application than Inmon's all- encompassing "Corporate Information Factory" approach. As for being the "father of data warehousing:" although Ralph Kimball's multidimensional database design schema is very popular for data marts and data warehouses, the honor for the title indisputably goes to Bill Inmon, as he was the one who made the term data warehouse popular.

Clay Rehm’s Answer: The major difference in the philosophies boils down to database design. Inmon favors third normal form design while Kimball favors star schemas (denormalized) design. What is interesting is that both will work. However you cannot walk into a client site, or your own company, and expect to push one philosophy. Each company and each data warehouse is different. In my experience, I always end up creating some hybrid approach that uses tools and techniques provided by Inmon, Kimball, myself and others. Regarding who the “father” of data warehousing. Who cares! You will do well to not take sides and use both of them as guides. The major difference in the philosophies boils down to database design. Inmon favors third normal