Virtual machine placement in cloudlet mesh

Abstract

In the era of Internet of Things (IoT), the concept of was proposed, which can be considered as a micro data center in a box deployed at the edge of the mobile networks. The integration of and wireless mesh networks gives rise to the idea of cloudlet which offers services to a large number of IoT applications simultaneously. With virtualized servers in each cloudlet, users issue networked VMs to run their applications, and then the broker or orchestrator of the mesh infrastructure decides the resource allocation. This process is known as the VM placement. With the network topology reconfigurability and the adaptive bandwidth provided by the Software Defined Networking (SDN) technologies, new opportunities arise to optimize the VM placement problem. In this paper, we address some of these opportunities. Firstly, we propose two optimal algorithms to determine the maximum number of accepted VMs into the mesh under both homogeneous and heterogeneous cases, respectively. Secondly, we design another two optimal algorithms to minimize the total inter-cloudlet communication traffic in both homogeneous and heterogeneous mesh, respectively. Thirdly, we study the VM placement problem under the multi-tenant scenario, which we prove to be NP-hard. A heuristic algorithm is proposed to give an efficient solution. The evaluation results validate the efficiency of our algorithm.