Given the title, one question that arises is how Hackney represents the relationship between the data mart and the enterprise data warehouse. The two natural choices are to work from the bottom up, assembling the data warehouse from the data mart, or from the top down, distributing subset departmental or subject area marts from the central warehouse. The author, and the industry, are astride the horns of this dilemma.

To his credit, Hackney navigates between these two major risks. If the incremental marts are built inconsistently and on an ad hoc basis, then you get what he calls legamarts--new data marts that are already legacy systems, obsolete and tightly coupled with an inflexible design. If the subset data marts presuppose the completion of the enterprise warehouse, then delivering the result is delayed by the time-consuming, expensive, and difficult task of completing the project across the entire enterprise.

In these times of accelerated delivery schedules and just-in-time everything, delays mean canceled projects. The author’s ingenious recommendation is to design a framework for the warehouse and implement the data marts incrementally within that framework. This solution is related to, but distinct from, both George Zagelow’s advice to drive out the inconsistencies on the way up and Ralph Kimball’s perspective, which says that if you do it right the first time, the business will grow and be enriched.

The first section, “Background and Evolution,” is a comprehensive treatment of basic semantic distinctions--OLTP and OLAP, ROLAP and MOLAP, DSS and EIS, facts and dimensions and aggregations, data scrubbing and metadata, the star schema and the operational data store.

The second section is a top-down discussion of “Preparation and Planning.” The author skillfully emphasizes a business-driven over a technology-driven approach; the spiral method of iterative development over the waterfall method; and the overriding importance of solving a problem that is causing pain to the organization in order to sustain the budgetary, political, and business will required to complete the undertaking.

The final section, on “Development and Implementation,” contains separate top-down discussions on the subset and incremental approaches to data marts. The description of the data mart corresponds to the prototype for the warehouse, with the departmental inconsistencies and semantic anomalies driven out or neutralized by the abstraction or imposition of an overriding architecture or framework.

The strengths of this text include its comprehensive and complete coverage; a wealth of detailed distinctions, which are not tied to any specific tools or vendors; and its treatment of metadata and business requirement–driven architecture (as opposed to mere technology).

Although the author commands a wealth of professional experience and knowledge, little of this experience is incorporated into case histories or clinical vignettes in the text itself. In a book of this length, one might reasonably expect more details on what is happening in the retail, publishing, or banking industries, even if the particular firms have to be disguised. This omission limits the audience to more technically inclined staff and managers. Still, the 20-page glossary and index will be useful to executives. (The bibliography is short and mostly limited to trade press magazine articles.)

Many acronyms are repeated excessively and unnecessarily in an attempt to give them a currency they are unlikely ever to attain. Finally, the author is not necessarily responsible for the redundant printing of diagrams. (For  example,  see pages 42 and 333, 50 and 373, 147 and 372, and 280 and 383, where identical pairs of graphics are printed.) The diagrams (and the entire text) are superbly produced in handsome grayscale with crisp, distinct, errorless graphics and text. Insofar as the purpose of the data mart and its senior relative, the data warehouse, is to reduce business uncertainty through knowledge, Hackney delivers both understanding and good counsel in the struggle for data integrity and access to usable information. This book is sure to be a helpful addition to the libraries of database administrators, data strategists, and knowledge managers.