A method and tools for large scale scenarios

Abstract

Formal scenarios have many uses in requirements engineering, validation, performance modeling, and test generation. Many tools and methodologies can handle scenarios when the number of steps (interleaved inputs and outputs of the target system) is reasonably small. However, scenario based techniques do not scale well with the number of steps, number of actors, and complexity of behaviors and system interactions to be specified in the scenario. First, it is impractically tedious and error-prone to specify thousands of input steps and corresponding expected outputs. Second, even if one can write down such large scale scenarios, confidence in their correctness is naturally low. Third, complex systems requiring large scale scenarios tend to require many such scenarios to adequately cover the behavior space. This paper describes the motivations for and problems of large scale scenarios, as well as the LSS method, which uses automated and semi-automated techniques in describing, maintaining, communicating, and using large scale scenarios in requirements engineering. The method is illustrated in two widely divergent application domains: military live training instrumentation and electronic mail servers. A case study demonstrates the practical and beneficial use of LSS in architectural modeling of a complex, real-world system design.