Although sound performance analysis theories and techniques exist, they are not widely used because they require extensive expertise in performance modeling and measurement. The overall goal of our work is to make performance modeling more accessible by automating much of the modeling effort. We have proposed a model interoperability framework that enables performance models to be automatically exchanged among modeling (and other) tools. The core of the framework is a set of model interchange formats (MIF): a common representation for data required by performance modeling tools. Our previous research developed a representation for system performance models (PMIF) and another for software performance models (S-PMIF), both based on the Queueing Network Modeling (QNM) paradigm. In order to manage the research scope and focus on model interoperability issues, the initial MIFs were limited to QNMs that can be solved by efficient, exact solution algorithms. The overall model interoperability approach has now been demonstrated to be viable. This paper broadens the scope of PMIF and S-PMIF to represent models that can be solved with additional methods such as analytical approximations or simulation solutions. It presents the extensions considered, describes the extended meta-models, and provides verification with examples and a case study.
Read full abstract