A Study of Factors Affecting Software Testing Performance and Computer Program Reliability Growth

摘要

Since the introduction of computing systems three decades ago, software development has grown increasingly in importance and complexity. Today the expense incurred by users in producing and maintaining programs exceeds ten billion dollars (Phister, 1976) and constitutes a large fraction of some corporate budgets (Dorn, 1978). Joint revenues of independent software suppliers exceeds one billion dollars, while revenues of computer manufacturers is approximately twice that (International Data Corporation, 1978). The increases in software-related expense far outstrip the growth in hardware costs. While the labor-intensive nature of software production has contributed to its increasing cost, competitive pressure and increased manufacturing automation have reduced the cost of hardware elements. It is for this and other reasons that Boehm (1973) predicted that software costs will constitute 90 percent of total system costs by 1985. As user needs become satisfied, software systems are designed to increasingly demanding specifications. This rapid growth in the number and complexity of desired systems demands equivalent increases in the number of qualified software specialists. More importantly, it dictates a maturity in software development methodology. The current state of 'software engineering' does not exhibit this degree of maturity, nor does any significant proportion of software workers employ those tools and techniques which most influence the quality of software systems. (Author)