Validation of Service Level Agreements Using Probabilistic Model Checking

With the fast growth of Information Technology (IT), organisations rely mostly on web services, cloud services and recently on Big Data Analytics services (BDA services), in order to support their business services. To securely use these services, service clients sign a Service Level Agreement (SLA) with service providers, regarding a particular service provision. Typically, SLAs define the properties that need to be preserved during the provision of a service (e.g., quality of service properties) and actions that will be applied if the service provision violates the defined properties (e.g., penalties or re-negotiation). Whilst significant research has focused on monitoring SLAs during the provision of services, the exploration and validation of the potential consequences of SLAs for the involved parties prior to putting them in operation is not addressed by existing research. In this paper, we present an approach to SLA validation that is based model checking. Our approach is based on the translation of SLAs expressed in WS-Agreement into models of the probabilistic model checker PRISM and the validation of SLA properties using the model checking capabilities of this tool.