A Prediction Model for the Probability of SLA Matching in Consumer Provider Contracting of Web Services

Abstract: Future e-business models will rely on electronic contracts which are agreed dynamically and adaptively by web services. Thus, the automatic negotiation of Service Level Agreements (SLAs) between consumers and providers is key for enabling service-based value chains. The process of finding appropriate providers for web services seems to be simple. Consumers contact several providers and take the provider which offers the best matching SLA. However, currently consumers are not able forecasting the probability of finding a matching provider for their requested SLA. So consumers contact several providers and check if their offers are matching. In case of continuing faults, on the one hand consumers may adapt their Service Level Objects (SLOs) of the required SLA or on the other hand simply accept offered SLAs of the contacted providers. By forecasting the probability of finding a matching provider, consumers could assess their chances of finding a provider offering the requested SLA. If a low probability is predicted, consumers can immediately adapt their SLOs or increase the numbers of providers to be contacted. Thus, this paper proposes an analytical forecast model, which allows consumers to get a realistic assessment of the probability to find matching providers. Additionally, we present an optimization algorithm based on the forecast results, which allows adapting the SLO parameter ranges in order to find at least one matching provider. Not only consumers, but also providers can use this forecast model to predict the prospective demand. So providers are able to assess the number of potential consumers based on their offers too. Justification of our approach is done by simulation of practical examples checking our theoretical findings.