Derivation of passage-time densities in PEPA models using ipc: the imperial PEPA compiler

Computer Performance Engineering

2011

Abstract. There have historically been two approaches to performance modelling. On the one hand, textual language-based formalisms such as stochastic process algebras provide portable and scalable solutions to compositional modelling. On the other hand, graphical formalisms such as stochastic Petri nets and stochastic activity networks provide an automaton-based view of the model, which may be easier to visualise, at the expense of portability. In this paper, we argue that we can achieve the benefits of both approaches by generating a graphical view of a stochastic process algebra model, which is synchronised with the textual representation, giving the user has two ways in which they can interact with the model. We present a tool, as part of the PEPA Eclipse Plug-in, that allows the components of models in the Performance Evaluation Process Algebra (PEPA) to be visualised in a graphical way. This also provides a natural interface for labelling states in the model, which integrates with our interface for specifying and model checking properties in the Continuous Stochastic Logic (CSL). We describe recent improvements to the tool in terms of usability and exploiting the visualisation framework, and discuss some of the general features of the implementation that could be used by other tools. We illustrate the tool using an example based on a model of a financial web-service application.

Section: Related Workmentioning

confidence: 99%

Visualisation for Stochastic Process Algebras: The Graphic Truth

Smith

Computer Performance Engineering

2011

“…As PEPA benefits from extensive software support, a number of analysis tools are readily available for re-use in this context. Here, each PEPA model is run through ipc [15]. It translates the description into a format suitable for hydra [16], which performs passage-time analysis and stores the results to disk.…”

Section: Analysis Tools In the Back-endmentioning

confidence: 99%

Service-Level Agreements for Service-Oriented Computing

Clark

Recent Trends in Algebraic Development Techniques

Tribastone

2009

Abstract. Service-oriented computing is dynamic. There may be many possible service instances available for binding, leading to uncertainty about where service requests will execute. We present a novel Markovian process calculus which allows the formal expression of uncertainty about binding as found in service-oriented computing. We show how to compute meaningful quantitative information about the quality of service provided in such a setting. These numerical results can be used to allow the expression of accurate service-level agreements about service-oriented computing.

“…For our model of the web service we are interested in the response-time as observed by a single user. Often we are interested in average response-time but compiling the PEPA models to CTMCs allows a finer grained analysis known as passage-time quantile analysis [12,13]. This allows the prediction of not just the average response-time but the response-time profile, such that we know the probability of receiving a response at or within any given time t after the request was made.…”

Section: Markovian Modelling With Many Modelsmentioning

confidence: 99%

Quantitative Analysis of Web Services Using SRMC

Clark

Formal Methods for Web Services

Tribastone

2009

Abstract. In this tutorial paper we present quantitative methods for analysing Web Services with the goal of understanding how they will perform under increased demand, or when asked to serve a larger pool of service subscribers. We use a process calculus called SRMC to model the service. We apply efficient analysis techniques to numerically evaluate our model. The process calculus and the numerical analysis are supported by a set of software tools which relieve the modeller of the burden of generating and evaluating a large family of related models. The methods are illustrated on a classical example of Web Service usage in a businessto-business scenario.