Data Management Insights

Data Warehouse Design

Author: Stephen Vandivier, President, Avanco International, Inc

Data warehouse applications have flourished as the realization has grown that data warehousing greatly facilitates evaluation of the enterprise and Just-In-Time information and decision making. Data warehouses, large electronic repositories of information accrued in a structured manner over time by an enterprise, provide a non-volatile single source reference point for information. Data warehousing answers the need in IT today for value-added applications that provide competitive advantage.

A Data Warehouse is a complete architecture, and as such, requires a rigorous, yet iterative design approach and development methodology to ensure successful deployment. This Insight will focus on architecture considerations, design principles, and a development methodology for data warehouse construction that we at Avanco have found to be indispensable. Following a solid "Data Driven" methodology can ensure a successful Data Warehouse project.

Data Warehousing

In contrast with the OLTP environment which is a volatile transactional data entry oriented environment, data warehouses are much more historical in nature, tend to be measured in terabytes of information, are organized and queried by subject rather than process, integrated within the enterprise, and are accessed by all levels within the organization. The huge volume of data and data sources inherent in the data warehouse necessitates a solid requirements oriented approach to its construction and a much different development methodology than the typical life cycle approach favored in building OLTP systems.

A fully functional data warehouse is an architected environment that typically summarizes data to the lowest common denominator or atomic level to provide the informational requirements of an organization. A data warehouse is an architecture comprised of many integrated enterprise components based on data and technology.

A data mart is a departmental instance of a data warehouse structured to meet the focused informational requirements of a distinct group within the organization. Organizations often impatiently embark on a data mart development effort as a low cost way to quickly get at their enterprise data. Data marts are often developed without considering an enterprise architecture approach. As a result, it is often difficult to draw many data marts together into a corporate data warehouse due to this lack of an enterprise strategy.

It is imperative that before embarking on either a departmental data mart or an enterprise data warehouse that the development team considers the proposed system and data architecture with an eye toward a fully corporate view of all data within the organization. The development methodology and standards to follow during design and development of either a data mart or data warehouse should be flexible enough to encompass the current or future needs of all departments in the enterprise.

To read more of this paper and to learn more about our unique Data Warehousing methodology and design techniques, email