Managing the final phases of software development using software reliability modeling
George D. Huensch, L.A. Bergen, Dennis A. Christenson, Akli Gana, Mario Garzia, D. Paul Smith · International Symposium on Switching · 2005
In the development of new releases of large software systems, the software is evaluated by various testing organizations toward the end of the software development interval. This paper de scribes the use of software reliability models and reports on metrics derivedfiom these models in managing the development resources in this "end game". Fault density and failure intensity are two metrics used both to monitor development process capabilities and also to measure customer satisfaction with the product. During the planning phase, targets are set based on customer expectations, to minimize faults remaining and the rate of failure occurrences at release. Reports are then generated regularly for the project managers showing the fault finding progress, the estimated number of faulls remaining, a 'rating' of each software component and estimates of current and future failure intensity. The "rating" of each component is based on the likelihood of reaching the fault density target at the release of the software system. The amount of time needed to reach the fault density and failure intensity targets, as well as the resources needed are provided by the reliability models. This information can then be used to allocate resources to those areas of the system where they would be most fruitful.