Optimal Cloud Broker Method for Cloud Selection in Mobile Inter-cloud Computing

Abstract

In the next generation of Cloud computing systems (so-called Inter-Cloud), it is expected that multiple Cloud Service Providers (CSPs) will collaborate or cooperate together to form the so-called Cloud Market, with the goal to advertise their services and associate prices to their end users. In this environment, mobile users will be able to better negotiate and select the CSP that better suits their budgetary and technical needs in a more flexible manner. However, despite the benefit of having multiple CSPs to choose from, mobile users will have to cope with several issues, for instance, that of selecting a suitable CSP to handle the incoming service request from the users. To address this problem, this paper proposes an optimal Cloud Broker. The role played by the optimal Cloud Broker is to provide the best possible match for mobile users and CSPs. The Cloud Broker design is formulated by using the framework of Semi-Markov Decision Process (SMDP). Considering the average cost criterion as the optimality criterion, the Value Iteration Algorithm is used to find the optimal Broker policy. Numerical results are presented, demonstrating the effectiveness of our proposed Broker design, in a scenario where a wireless service provider has its own Cloudlet and a Service level Agreement (SLA) has been established with a public CSP to handle the overload occurrences.