Abstract

The systems industry has now experienced al most three decades of growth and devel opment. In that period, a 1 arge number of analysis tools and techniques have been proposed to aid the development process. Early systems were supported by analysis techniques which had been used for some time in precomputer systems. Next, the precomputer techniques 'were modified to meet some of the unique requirements of computer based systems. Succeeding generations of analysis tools continued to provide improved support to the analysis process. In recent years, a series of structured analysis tools and techniques has been introduced to the industry. At this point, a large number of competing analysis techniques exist and are widely used. However, they are not cl early understood by many practici ng professi onal s. They tend to be i ncompl ete, requi ri ng careful eval uation and integration to result in coherent analysis processes. Unfortunately, the 1 iterature on the subject tends to concentrate on the strengths of individual tool s, often implying that a single analysis process can address all needs. In reality, all analysis tools and techniques are incompl ete. While specific approaches provide support for specific analysis problems, none cover all of the system issues of interest. Traditional techniques tended to provide good detail on input and output detail. In addition, traditional analysis approaches cl arified flows of information through the organization. Later approaches considered data storage and provided tool s to represent procedural system aspects. Structured techniques concentrate on the structure of data fl ows, data, and control . Unfortunately, modern analysis approaches exhibit improvements in some areas of analysis while neglecting some of the strengths of older techniques. This paper presents a comparative examination of analysis techniques to aid practicing professionals in the choice of tool s for devel opment efforts. The comparison i s supported by a set of dimensions which represent the various system aspects of interest during analysis. These dimensions include considerations of system structure, functions, procedure, input detail, output detail, and mechanisms responsible for functions. In addition, analysis techniques may be compared in terms of their ability to support high and low level analysis and to support effective communication between systems professionals and their customers. The comparison of analysis techniques clearly shows that traditional approaches failed to consider system structure i ssues. However, modern tools fail to consi der some of the traditi onal issues of interest. For exampl e, most of the , structured analysis methods fail to provide any support for I/0 detail. In addition, almost all of the currently popul ar analysis techniques assume that all functions will be implemented in software. Only SADT and some of the older techniques support the analysis of mechanisms responsible for functions. Current man-machine concerns make mechanism analysis critical. The comparison of techniques indicates a need for the combination of multiple tools to provide complete coverage of the issues of interest during analysis. In the comparison, the strongest approaches were those which explicitly required the use of multiple tools. For exampl e, HIPO is a package which is quite compl ete, despite its age. The comparison process provides sufficient detail to support the choice of techniques which can be combined into complete packages.

Share

COinS