Continue in 2 seconds

Building Successful Data Warehouses

Published
  • November 01 1999, 1:00am EST

This month we examine some of the neglected aspects of data warehousing that cause companies to fail to achieve a satisfactory return on investment.

The greatest cause of data warehouse failure to provide a reasonable return on investment is the building of data marts that are based on separate, disintegrated and redundant data versions. Such data marts have been attractive, as they are easy and quick to build. The rationale for these data marts has often been: Don't build a data warehouse. It takes too long. Other parts of the business can't decide how they want to use "our" data, and who cares. We know our own data. The result: data marts that perpetuate the disintegrated data sources as disintegrated information that has no enterprise credibility.

In contrast, when redundant data versions are integrated, all processes share access to the same version of that common data. Whenever any one process adds to or changes that data, all processes can use the result of that addition or data change immediately. Information derived from integrated databases is, therefore, consistent and accurate. Data warehouses that are built to deliver this information are successful and have a higher return on investment.

Strategic Information Systems Plan

The corollary to this question is: How can enterprises maximize the return on investment from data warehouses? The answer is now obvious. As an architect develops an overall design for a building before construction begins, so also an enterprise should develop an overall design for a data warehouse before its construction begins. This does not take long ­ typically only four weeks to develop a strategic model of the enterprise and document it in a strategic information systems plan (SISP).

The SISP report clearly indicates how to build an integrated data warehouse, quickly. This is based on the strategic, tactical and operational business plans that define the information that managers need to run the business. The project plans documented in the SISP report define how priority data marts can be implemented progressively to satisfy urgent business needs. As these separate data marts are delivered, they fit together precisely, evolving into an integrated data warehouse without the problems of redundant, disintegrated data and information discussed earlier. This is the same way that complex buildings are built in the construction industry: from an overall, integrated plan ­ with floors and other parts of the building progressively built to fit together precisely.

An example of a typical SISP report can be read online or downloaded from the IES Web site by clicking on the Projects link from any page at ­ http://www.ies.aust.com/~ieinfo/. This was developed for a regional bank in South Korea, Kwangju Bank. It is in English and was written for a business and IT audience. The approach described in this SISP report is applicable to any organization, government or commercial, and any industry.

Successful data warehouses are built through a design partnership between business experts (who know the business) and IT experts (who know computers). Business experts can learn and apply data modeling and normalization principles as well as, and sometimes better than, many IT experts. They know the business intimately while we may not. But they need to use a variant ­ called business data modeling.

In the March issue of DM Review, I discussed how a banker used a strategic model of his bank to identify competitive opportunities for global Internet banking. He had been trained in business data modeling. We both talked about the business advantages and threats and considered the technology implications. The bank was in Korea, so he was speaking only Korean. I was responding only in English. Yet we understood each other precisely, as we both used the data model to communicate! It is a powerful communication medium.

A Certified Business Data Modeler (CBDM) course series has been developed to assist you, specifically for self-study use by business staff and IT staff. It includes two courses: Data Modeling Concepts and Business Normalization Concepts; and a Data Modeling Case Study Workshop. This workshop enables business and IT staff to apply the skills they have learned in the concepts courses to a real-life case study, using an I-CASE modeling tool supplied as part of the CBDM course series.

The courses can be taken at work or at home ­ at any convenient time or place. Three delivery options are available: PowerPoint, Internet/intranet or classroom. You can read the course outlines and view the introductory slides online by going to the IES Web site athttp://www.ies.aust.com/~ieinfo/ and clicking on the Courses link.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access