Transformations in PEPA Models and Stochastic Probe Placement
Allan ClarkStephen Gilmore · 2009
We describe a language of transformations over performance models written in PEPA [1]. This language allows many similar models to be derived from a single input model. This in turn allows experimentation over a variety of similar configurations of a modelled system. We then turn our attention to the use of transformations as a probe placement language. A stochastic probe is an observational component which may be attached to the entire system as a global observer or to a sub-portion of the model as a local observer. Local observers allow the calculation of such analyses as the response-time of a service as observed by a single client. We use our transformation language as a way for the user to describe where in the system a local probe or probes should be placed.