PUTTING THE DSS TOGETHER _

6.12 PUTTING THE DSS TOGETHER _

In Chapter 3, we described a typical Web-based DSS architecture (Figure 3.1). The structure consisted of several Web servers and a client, all integrated to help decision- makers in business intelligence work. Integration is still a difficult issue, and more crit- ical now that DSS/BI tools have migrated to the Web and also readily provide collabo- ration and communication tools to decision-makers. These and other considerations must be handled as new DSS are developed and legacy systems migrate to the Web.

Development tools increase the productivity of developers and help them con- struct a DSS responsive to users' needs. The philosophy of development tools and gen- erators is based on two simple yet very important concepts: the use of highly auto- mated tools throughout the development process, and the use of prefabricated pieces in the manufacturing of a whole system whenever possible (e.g., component reuse)

(Yongbeom and Stohr, 1998). The first concept increases the productivity of the devel- oper in the same way that an electric saw improves the productivity of a carpenter who formerly used a hand saw. The second concept increases productivity analogously to the way a prefabricated wall increases the productivity of a carpenter building a house. Fortunately, a component is not "consumed" when it is used, in the sense that the wall is. It can be used again and again. As the components of a DSS are developed, care must be taken to make them fit together (like the components of a house—the plumb- ing must fit inside the walls but must link the outside water supply to the sinks and tubs, and so on).

A DSS is more than just the DBMS, MBMS, user interface, and knowledge com- ponent. There are interfaces among the components and with external systems.

C H A P T E R 6 DECISION SUPPORT SYSTEM DEVELOPMENT

Typically, DSS databases must be refreshed regularly from other source databases. There may be special tools for necessary functions like report generation. There may

be several databases and models, each of which is developed and used differently; and there may be many people involved in the development in terms of data gathering (refer to the Web Chapter on P&G's supply chain redesign). Not only do the compo-

nents have to be constructed, but the specific tools and generators for development also must be selected, installed, and managed.

The system core includes a development language or a DSS generator. Some of the necessary capabilities mentioned above are integrated into DSS generators. Others can be added as needed. These components can be used to build a new DSS or update

an existing one. The construction involves the combining of software modules. Fortunately, the newer object-oriented operating systems provide a consistent, user-

friendly environment for DSS development. Tools and generators that run in them can easily share results and data. Since a consistent, user-friendly interface can be devel- oped quickly (say, in Microsoft .NET), component interfacing problems are generally minimal. In fact, Web browser GUI interfaces are commonly used to front-end legacy DSS and databases instead of rewriting the whole system (see DSS in Action 3.4). Alternatively, legacy systems have been moved to Web servers along with creating the Web-browser front end.

TRENDS IN DSS/BI IMPLEMENTATION In Chapters 2 and 3, we described some recent developments in DSS/BI.There are sev-

eral trends that continue to impact DSS/BI applications. They include: • Managers are more readily accepting DSS/BI tools, techniques, and methods.

• Artificial intelligence tools and methods (expert systems, neural networks, genetic algorithms, fuzzy logic, etc.) are being embedded in DSS/BI. • Web technologies continue to enable new developments in DSS/BI from data, information, and knowledge access to direct communication and collaboration. • GSS continues to proliferate through collaborative computing. • Computer technology continues its fast-paced evolution. Capabilities are increas-

ing dramatically, and costs are decreasing. This leads to greater capabilities being

embedded in DSS/BI. • Enterprise resource management/enterprise resource planning (ERM/ERP) sys- tems, though extremely expensive, are proliferating. These often provide and incorporate DSS methods for improved decision-making. One future develop-

ment of the IMERYS model (Case Application 6.1) will be to use its results to

drive a new ERP.

• CHAPTER HIGHLIGHTS • The traditional system development life cycle (SDLC)

• There is a need for good project management skills in is a structured approach for managing the development

system development team leaders. of information systems.

• In practice most information systems do not succeed. It • The four fundamental phases of the traditional SDLC

is important to understand the factors that lead to are planning, analysis, design, and implementation.

failure so that they can be recognized early. • Each phase of the SDLC has several small steps, each

• DSS are usually developed by prototyping (iterative with its own techniques and deliverables.

design, evolutionary development) development • Computer-aided software engineering (CASE) tools are

methodology.

useful for managing large information-system • Prototyping is a rapid application development ( R A D )

P A R T II DECISION SUPPORT SYSTEMS

• Prototyping consists of rapid cycles through the • Selecting DSS software and hardware is difficult fundamental phases of the SDLC, with user feedback

because it involves both quantitative and qualitative guiding system modifications. This is a form of joint

factors.

application development (JAD) and rapid application • There are many Web-based DSS tools and generators development (RAD). Typically DSS developed with

on the market. The appropriate ones for building a prototyping continue to evolve following deployment.

specific DSS must be selected carefully. • Iterative prototyping methodology is most common in

• DSS can be built by teams or by individuals. DSS development because information requirements

• A team building a DSS must follow a structured are not precisely known at the beginning of the process.

process.

• Prototyping helps the user understand the decision- • Most end-user DSS developed with an integrated tool making situation as the system evolves.

like an Excel spreadsheet are used for personal support. • New agile development and extreme programming

• The major benefits of end users developing their own methods are formal prototyping methods that are useful

DSS are short delivery time, users' familiarity with their for developing small- to medium-sized systems quickly.

needs, low cost, and easier implementation. • Change management is difficult; a number of

• End user-developed DSS can be of poor quality. organizational factors must be present if it is to occur

Appropriate controls based on system-development successfully.

methods can improve quality. The two primary ones are • The deceptively simple Lewin-Schein model of change

(1) to understand the model of the problem, and (2) to management (unfreeze, move, refreeze) elegantly

review the model carefully.

describes how change can be managed. • Assembling a DSS can involve many components and • DSS technology levels are DSS primary tools, DSS

their interfaces.

integrated tools (generators, engines), and specific DSS. • New DSS/BI software are incorporating artificial • DSS are typically constructed with a DSS generator

intelligence methods and better collaboration systems consisting of an integrated set of development tools.

(GSS), and exploit Web technologies.

• KEY W O R D S • DSS generator (engine)

• specific DSS (application) • DSS integrated tool

• evolutionary development

• system development life cycle • DSS primary tools

• feasibility study

• iterative design

(SDLC)

• DSS technology levels

• team-developed DSS • DSS tools

• phased development

• technology levels (of DSS) • end-user computing

• prototyping

• throwaway prototyping • end-user development

• rapid application development

(RAD)

« user-developed DSS