Author: Bill Inmon
Publisher: Wiley, 1999 (ISBN: 0-471-32888-X)
Introduction
Have you ever wondered what an Operational Data Store is? How do you know if you need one? What needs does it satisfy? This book explains the operational data store — its purposes, types,
design, and management. The book is a mix of theory, practice, and methodology. There are lots of big ideas in the book, and what you find may surprise you.
The Operational Data Store
The first chapter in the book explains why organizations need a data warehouse and, perhaps, an operational data store. The operational data store is a hybrid architectural construct. It generally
serves some operational needs as well as some decision support needs. As a hybrid, its design must necessarily be constrained by conflicting requirements. Some key reasons for building an ODS are
to integrate operational data, to use for decision support, and to stage data for analytical processing.
As in many situations, there are two main options: build or buy. There are several well-known packages that implement an operational data store. A later chapter discuss four types of operational
data stores, which are primarily distinguished by the speed at which data is read from operational systems and loaded into the ODS.
The Corporate Information Factory
The second chapter elaborates on another big Inmon idea—the Corporate Information Factory. The Corporate Information Factory is a framework for understanding the data processing and
architectural components of an enterprise. This chapter outlines the main constructs in the Corporate Information Factory and shows how the ODS fits into the overall environment.
Methodology
The heart of the book addresses how to build an operational data store. It includes chapters on high-level design, low-level design, hardware and software requirements, and staffing requirements.
Appendix A provides a work breakdown structure for an ODS project—a list of the major tasks required to build an operational data store. One chapter provides roles and a sample organization
chart for the ODS organization. Other chapters address standards, reengineering, data normalization, and the role of star schemas in the ODS.
Technology
The hardware and software requirements chapter addresses a number of interesting technologies that may be applicable to the ODS. Topics include massively parallel processing (MPP),
transaction-processing monitors (TPM), mixed workloads, distributed schema and routing, and the placement of data.
Managing the Operational Data Store
Two chapters are devoted to managing the ODS. One concentrates on the technical aspects, while the other concentrates on the people aspects. Technical topics include monitoring system utilization,
monitoring the growth of the user base, the daily and nightly workload, ways to distribute the workload, capacity planning and service level agreements.
The people-oriented chapter concentrates on the skills necessary to implement and manage the ODS, including requirements gathering, design, project management, development, testing, and system
integration. These skills correspond to the roles of project manager, DBA, data modeling/administration, development, metadata specialist, and tester. This chapter also recommends staffing levels
for development an maintenance of the ODS.
The Rest
One chapter provides ODS case studies. Appendix D provides many articles of interest on data warehousing. You can find many of these articles on the Data Management Review web site
(www.dmreview.com). Appendix C is a glossary, and Appendix B is a list of ODS project milestones.
Conclusions
Bill Inmon generates a lot of ideas in his writings, and there are two fundamental tendencies that characterize his work. The first is a historical approach. He addresses where we are in terms of
what has gone on before. The second tendency is an architectural approach. He addresses his topic in an architectural framework—something that few other authors do. Both of these tendencies
enrich the topic that he is addressing, in this case, the Operational Data Store.
The ODS is here to stay, and it’s becoming a more recognizable architectural construct. The ODS is a hybrid beast—partly in the world of operational systems and partly in the world of
decision support. This book will show where it fits in the corporate information factory as well as how to design it, build it, and manage it.
