Techniques for the construction and analysis of algebraic performance models
Graham Clark · ERA · 2000
The performance modeller may attempt to quantitatively analyse the behaviour of computer systems by building performance models. Such models may become unwieldy, and so high-level structured modelling techniques have been developed. A stochastic process algebra (SPA) provides such a technique, a compositional modelling calculus. Hillston's PEPA is an SPA, a classical process algebra enhanced to represent the performance of systems. This thesis uses PEPA as a foundation, and examines di#erent ways to assist the SPA performance modeller. A weak stage in the SPA methodology is the calculation of concrete performance measures, since much research does not focus beyond a steady-state probability vector. A framework is developed for specifying steady-state performance measures for PEPA models. The technique is used at the high-level of the process algebra, and not applied directly to states, or the stochastic process. It employs an enhanced modal logic to allow the modeller to identify inte...