February 5, 1999

MEMORANDUM

TO: NSF Design and Manufacturing Conference Attendees

SUBJECT: Scalable Enterprise Systems

We in the Engineering Directorate of NSF are pleased to open a discussion on a research area that we consider very promising. The primary objective of this research area would be to foster the development of a science base for enterprise-wide business automation. While the underlying concepts for this topic originated in the context of manufacturing systems, the application scope of the research is meant to include other functions of an enterprise system such as financial management, human resource management, and sales and marketing. Because the Internet looms large as a deployment environment, issues of scalability will be crucial in the development of this science base.

From its inception, the commercial software market has been driven by the needs of automating business functions. In recent years the fastest growing part of the market has been ERP (enterprise resource planning) systems. Some companies specializing in this sector have enjoyed annual growth rates of 50% or more and have been the darlings of Wall Street.

The concept of a consistent suite of interoperable application programs to serve all major functions of a business enterprise had its origin in manufacturing. Those engaged in designing CIM (computer integrated manufacturing) systems soon realized that such systems could not function in isolation from other major functions of a manufacturing enterprise, such as cost accounting, human resource management, sales and marketing, and purchasing. These considerations led to the development of concepts such as "MRP (materials requirements planning)," "enterprise level systems," "supply and value chains," and others which, in turn, gave rise to ERP systems. Indeed, manufacturing has been the source of some of the best models and science-based concepts that underlie today’s ERP systems, and it remains an important component in all ERP suites. More than ever, a science base to underpin the entire field of enterprise-wide business automation is needed now, and the manufacturing community should take the lead.

We note, in passing, that enterprise level systems are often viewed in terms of their three principal layers: production layer which consists mainly of manufacturing, back-office layer which includes such core functions as human resource management and accounting and front-office layer which comprises the customer and inter-business functions such as sales and marketing.

Current ERP products represent an evolutionary development in commercial data processing. There is little if any theoretical basis for this development. What theory that exists lies in the database portion of such systems and in optimization problems for operational functions, such as resource planning and scheduling. We note that "theory" in this context should be interpreted modestly. It is mainly a descriptive theory comprising concepts that simplify, clarify, and unify. Such a theory is valued for its semantic power to express and clarify, rather than its power to predict.

The promise of ERP is more than enterprise-wide interoperability and consistency. Its objectives should also include standardization of its principal functional modules to minimize customization. This objective is far from being achieved. ERP implementations require a high degree of customization. Full implementation often takes years with many efforts never completed. What is needed is a theoretical foundation similar to the one provided to database systems by the relational model. Without such a foundation it is not surprising that the promise of ERP systems has proved elusive.

The term scalability refers to Internet related issues. Internet has become a universal medium for enterprise level software deployment. Whether the network is the public Internet or some private variant thereof, the operating environment now greatly stretches the range of scalability, from a few users to millions of simultaneous users. This is true not only of consumer oriented retail operations on the Internet, but also business-to-business e-commerce and deployment of enterprise-wide systems. In this context, scalability should mean "no change in software and no degradation in efficiency" as the number of users increase indefinitely.

We believe that scalable enterprise systems represent a natural extension of the enterprise level software environment created by the manufacturing community. While the work on enterprise-level systems led to the development of a fast growing sector in the commercial software market, many of these developments have been largely ad hoc. We believe that the principal objectives of enterprise level integration cannot be achieved without developing a science base for business automation. To do so, in the context of the Internet-based deployment environment, is the aim of the research activity being considered.

In order to identify the basic research issues in this area, we are in the process of planning a workshop on this subject to be held in 1999. If you should have any questions or comments on this topic, please feel free to provide your input by e-mail (lmartinv@nsf.gov).

Eugene Wong
Assistant Director of NSF for Engineering

Louis A. Martin-Vega
Director, Division of Design, Manufacture
and Industrial Innovation (DMII)