From uml to SRN: A tool based support for performability modeling of distributed system considering reusable software components

The analysis of the system behavior from the pure performance viewpoint tends to be optimistic since it ignores failure and repair behavior of the system components. On the other hand, pure dependability analysis tends to be too conservative since performance considerations are not taken into account. The ideal way is to conduct the modeling of performance and dependability behavior of the distributed system jointly for assessing the anticipated system performance in the presence of system components failure and recovery. However designing and evaluation of the combined model of a distributed system for performance & dependability analysis jointly is burdensome and challenging. Focusing on the above contemplation, we introduce a framework to provide tool based support for performability modeling of a distributed software system by proposing an automated transformation process from high level UML (Unified Modeling Language) notation to SRN (Stochastic Reward Net) model and solves the model for early assessment of software performability. UML provides enhanced architectural modeling capabilities, but it is not a formal language and does not convey formal semantics or syntax. That?s why we present the precise semantics of UML models by formalizing the concept in the temporal logic cTLA (Compositional Temporal Logic of Actions). cTLA describes various forms of actions through an assortment of operators & techniques which fit excellently with UML models applied here and cTLA also provides the support for incremental model checking. The applicability of our framework is demonstrated in the context of performability modeling of a distributed system to show the deviation in system performance against failure of system components.