Statistical modeling applied to managing global 5ESS®-2000 switch software development
Akli Gana, Steel T. Huang · Bell Labs Technical Journal · 2002
In the development of new releases of large software systems, the software is evaluated by various development and testing organizations. To ensure that software faults causing customer-affecting failures are minimized at each stage of the development cycle, we use developer-oriented and customer-oriented metrics to measure software quality. Developer-oriented metrics focus on faults and fault densities during testing to monitor development process capabilities and software quality. These metrics are expressed as the number of faults per thousand of non-commentary source lines of code (KNCSL). Customer-oriented metrics focus on failure intensity during software execution, thus providing a user's view of the software quality. These metrics are expressed as the number of failures per unit of time. This paper describes the use of empirical software tracking models and software reliability models, as well as reports derived from these models, in managing the 5ESS®-2000 switch development resources and product quality.