SLA and cost-aware cloud service provisioning through matching game and TOPSIS model

Abstract

Cloud service delivery is based upon Service Level Agreements (SLA) reflecting a customer and service provider signature agreement setting out the terms of the agreement, Providers must use resources efficiently to minimize the costs of provisioning services. Therefore, strategies are required that take into account multiple SLA parameters and efficient resource allocation. Recent work takes different strategies with single SLA parameters into consideration. These approaches are, however, restricted to simple workflows and to single tasks. The preparation and execution of service requests that take into account several SLA parameters such as required amount of CPU, storage, memory and price are still unresolved research challenges. In our studies we have found matching game approach can be useful in dealing with multiple criteria of cloud resources which can be allocated fairly and efficiently. The emphasis of this paper is the implementation of an SLA provisioning algorithm for game theoretical resources that takes into account the user equality and the use of resources for both. To rank the clients on the basis of their requirements against the resources provided by the clouds service and cloud services against the requirements of clients we have used TOPSIS algorithm. In our thesis, the TOPSIS implementation tests on a 4-VM cluster demonstrate how best this algorithm can be managed in a fair comparison with the Jain’s Fairness Index analysis.