In this paper, we present an approach to modeling and analyzing the quality of service (QoS) characteristics of compositional cloud services in a federated ecosystem. Our approach first synthesizes service compositions of federated services into a small number of primitive models, basic elements and associative characteristics as common patterns, each of which has an analytical model of its QoS characteristics. A cloud-based federated system is defined through recursive composition of the primitive models and thus can be modeled and analyzed through a decomposition and aggregation process based on the QoS characteristics of primitive models. The main objective of our approach is to support the design of cloud-based federated ecosystems and the management of service level agreements for business applications in such systems. We examine cases and conditions within which a composite model's QoS characteristics have analytical solutions as a composition of the QoS characteristics of primitive models and those that may not have an analytical solution and therefore simulation-based solutions are required. Test results are discussed for comparison of analytical and simulation approaches.
展开▼
