VMUAllocator e CBHSelection : novas políticas verde de alocação e seleção de máquinas virtuais em um ambiente de Datacenter

Abstract

Cloud Computing consists of providing computational resources, in which database services, storage, network and various applications through the internet stand out. Resources are provided through providers, made up of several server machines, forming an infrastructure known as a Data center. The entities that consume these resources have virtual machines distributed across all server machines in a Data center. The growing increase in Data center infrastructures has led to a large increase in electricity consumption. For this reason, several allocation policies have been proposed, because these policies aim at a more optimized allocation, in order to provide a greater reduction in energy consumption. This work presents a new Allocation Policy (VMUAllocator) whose main objective is to form pairs of virtual machines that can be allocated according to the maximum capacity of the servers, and a new selection policy (CBHSelection) which analyzes the virtual machines and classifies as "potentially problematic"machines, that is, that cause race conditions with other machines within the server. The proposed allocation policy demonstrated good results, standing out as the best energy policy in most environments. The proposed selection policy obtained good results in relation to energy consumption, coming in second place in all simulated environments.