banner image for William G. Smith & Associates


Many enterprises today are building one or more "data warehouses" as a means of dealing with the underlying data mess resulting from their disintegrated application systems and databases. Although the building of a data warehouse will not remedy the underlying systems and data mess, it can be an effective interim strategy to allow the appearance of consistent and integrated data, and provide a data store where different business processes can at least share the same data at the same time.

The principles of correct data organization are every bit as applicable (if not more so) to a data warehouse as they are to designing any shared database. In fact, since data warehouses tend to be searched and accessed in highly spontaneous and ad hoc ways, having a design which is completely sensible to the user is crucial - a mirror image of the real world persons, places, things, concepts and events about which the enterprise must collect and store data. This seminar combines the discipline of good data modeling practice with the additional concerns of identifying the best source data in existing systems, and planning the continuous or periodic movement of data from the source files/databases to the data warehouse.

Data Modeling for Data Warehouse is a skill-building seminar which teaches the familiar three-level approach to data modeling (Conceptual, Logical and Physical) applied to the problem of designing a data warehouse. It integrates the disciplines of entity/relationship modeling, data element analysis and standardization, normalization, and physical database design to fully equip the student to design a stable, flexible, highly sharable, effective, non-redundant data warehouse. Also, the necessary metadata to support the data warehouse (where did this data come from?) is addressed in detail with models and discussion of a repository in which to hold the appropriate metadata. The seminar material is completely independent of any particular vendor's CASE tool or physical data base management system (DBMS).

Because of the large amount of material covered, this seminar is appropriate for experienced data modelers who are/will be engaged in the modeling and construction of data warehouse. It focuses on the special and often confusing issues and differences between a data model independent of the existing systems/data, and a data model for a data warehouse which is constrained by the legacy data/systems which already exist in the business and which will serve as the sources of the data which will be imported into the data warehouse.


DURATION: 5 days

TARGET AUDIENCES: (recommended maximum number of attendees - 25)

PREREQUISITES: (very strongly recommended, not mandatory)