Toward green cloud computing environment by means of virtual machine consolidation

Abstract

The urge for computing resources has popularized the usage of cloud computing. Through the use of virtualization, cloud data center managers can reduce the Total Cost of Ownership and improve resource utilization. Even though several costs can be reduced through virtualization, the power consumption required from the whole data center ecosystem accounted for a substantial amount of all electricity consumed in the world. This also entails high levels of CO2 emissions that, together with the energy consumption, must be reduced to achieve green computing. This problem can be mitigated by packing virtual machines into a fewer number of physical machines to shut down idle ones. However, the usage of virtual machine consolidation to reduce energy consumption might affect the Service Level Objectives when dealing with the energy and performance tradeoff. This work has two main contributions. The first contribution is a Systematic Literature Review, which aims to provide a comprehensive understanding of virtual machine consolidation through a quantitative and qualitative analysis of several papers from the literature. The methodology and tools used are described to enable the reproduction of the results obtained. Regarding the second contribution, this work studies the use of a Mixed Integer Linear Programming model during the virtual machine placement step of consolidation. The model is used to provide a new mapping between virtual and physical machines that minimizes the energy consumption and the total number of migrations. The experiments were simulated using real CPU utilization traces and the models were solved using two commercial solvers. The results show that the proposed model on one of the solvers, in general, issues fewer migrations to the simulator. However, since several migrations issued to the simulator are feasible, there is an increase in energy consumption and performance degradation due to the migration overhead.