CAVMS: Application-Aware Cloudlet Adaption and VM Selection Framework for Multicloudlet Environment

Abstract

The mobile users offload the application to nearby cloudlet servers instead of the remote cloud for better end-user experience. Each cloudlet is able to process real-time applications with the help of virtual machines (VM). While multiple applications running on the cloudlet, the possibility of overprovisioning issue is unavoidable due to massive task-offloading requests from mobile devices. In this regard, balancing the load, among the cloudlets in a high-interactive applications scenario, is a promising issue. In order to balance the cloudlet load, migration of VMs from an overloaded cloudlet to an underloaded cloudlet is a favored solution. During this process, a well-designed migration mechanism must be outlined that can perform two steps such as VM selection and cloudlet adaption. In this article, an application-aware cloudlet adaption and VM selection framework has been devised for balancing the load in a multicloudlet environment. The candidate-cloudlet adaption is based on a migration efficiency indicator that reduces the response time and enhances load-balancing rate. Furthermore, the effectiveness of the framework has been evaluated by comparing with other state-of-the-art cloudlet-selection strategies.