A Virtual Machine Migration Policy Based on Multiple Attribute Decision in Vehicular Cloud Scenario

Abstract

Virtual Machines (VMs) offer great flexibility, reusability, and portability to manage applications in physical resources of a Cloud data center or a mobile cloud. These VMs may suffer a migration from one datacenter to another or from one cloudlet to another, due to the characteristics of vehicular cloud, such as high mobility and the need to meet the demands of a particular service. In this way, establishing a VM migration policy in a vehicular cloud becomes a challenge, since it has to deal with the unique challenges of vehicular networks, as well as to meet the demands of users' services without virtual machine migration, which impacts on the performance of such service. In this paper, we propose a decision policy based on multiple attributes to migrate VM in a vehicular cloud scenario. In this way, the proposed policy allows to decide more quickly if a VM migration should be carried out and to which cloudlet this machine should be migrated. Simulation results showed that the proposed policy reduced in 2% the amount of VM migration in the network, decreased in 3% the blockages of the migration requests, as well as a reduction in the inference time of approximately 5 ms.