A probabilistic measure of the delay incurred in an action of the system. Rates are drawn from the exponential distribution (i.e., rate r is the parameter of F(t)=1–e-rt). More specifically, if action is associated to rate r then that action will be executed within time t with probability 1–e-rt.
Published in Chapter:
Analysing Time-Related Properties of Service-Oriented Systems
Laura Bocchi (University of Leicester, UK), José Fiadeiro (University of Leicester, UK), Monika Solanki (University of Leicester, UK), Stephen Gilmore (The University of Edinburgh, UK), João Abreu (Altitude Software, Portugal), and Vishnu Vankayala (Lapilluz Software Solutions, India)
Copyright: © 2012
|Pages: 25
DOI: 10.4018/978-1-61350-432-1.ch003
Abstract
We present a formal approach for expressing and analysing time-related properties of service-oriented systems. Our aim is to make it possible for analysts to determine, based on models of services developed at early stages of design, what quality-of-service properties can be expected from, or offered by, the providers of those services. Our approach is based on an extension of SRML, a high-level modelling language developed in the Sensoria project for architectural and behavioural specification of dynamically reconfigurable service-oriented systems. The proposed language extension offers primitives that capture several kinds of delays that may occur during service provision. Quantitative analysis is supported by mapping SRML models to PEPA, a Markovian process algebra supported by a range of efficient software tools that can either confirm that required properties are met or provide feedback that can be used to improve the SRML model.