Lookup NU author(s): Rouaa Yassin Kassab,
Professor Aad van Moorsel
Full text is not currently available for this publication.
SLA (service level agreement) specification languages are designed to express monitorable contracts between a provider and a consumer of a service. It is of interest to determine if one can derive predictive models, e.g., a discreteevent stochastic model, for SLAs expressed in such a language, if possible in automated fashion. For this purpose, we study in this paper the mapping of the Web Service Level Agreement (WSLA), an XML-based SLA specification language tailored to web services, into reward metrics defined in the Stochastic Discrete Event Systems (SDES) formalism. We associate a formal semantics with WSLA elements and map these on SDES through a five step mapping process, which includes expressions for the metrics and functions on these metrics, the instances of time at which to predict, and the ultimate SLO compliance probability.The work in this paper provides the foundation for automation of the mapping, and we will identify precisely what input a modeler needs to provide to complement the automatable part.We illustrate the use of our approach through a simple case study of a stock quote service.
Author(s): Yassin Kassab R; van Moorsel A
Publication type: Report
Series Title: School of Computing Science Technical Report Series
Source Publication Date: April 2011
Report Number: 1245
Institution: School of Computing Science, University of Newcastle upon Tyne
Place Published: Newcastle upon Tyne