A method for dependability evaluation of software architectures

作者：Sedaghatbaf Ali; Azgomi Mohammad Abdollahi^*

来源：Computing, 2018, 100(2): 119-150.

DOI：10.1007/s00607-017-0568-3

• Software architecture
• Dependability evaluation
• Stochastic activity networks (SANs)
• Activity-marking oriented reward structures

Dependability is an important quality attribute for today's software systems, and the decisions made at the architectural design stage play a key role in the dependability of software products. Architecture-level dependability evaluation techniques are effective only if they support a practical architecture modelling notation and consider the impact of execution environment, error propagation and usage profile on dependability. Regarding these factors, the objective of this paper is to introduce a new method for evaluating the dependability of software architectures. UML is the de facto standard for modelling software architectures, and DAM is the most comprehensive UML profile supporting dependability annotations. In this paper, UML-DAM models are transformed into a hierarchical and coloured extension of stochastic activity networks and the dependability measures are quantified using the activity-marking oriented reward structures. The transformation and evaluation processes are detailed and implemented in a software tool. Also, to validate the applicability of the method, a case study is presented.

• 出版日期2018-2