Empirical Comparison of Power-efficient Virtual Machine Assignment Algorithms

The advent of cloud computing has changed the way many companies do computation, allowing them to outsource it to the cloud. This has given origin to a new kind of business, the cloud providers, which run large datacenters. In order to be competitive, cloud providers must keep the energy consumed by the datacenter low. 

One way to achieve this is with smart task assignment algorithms, which decide where tasks are to be placed upon their arrival. In this paper we compare the performance of multiple task assignment algorithms for saving energy. We assume tasks are in fact virtual machines that have to be assigned to physical machines, and we assume that the physical machines have a power consumption that increases superlinearly with the load. In particular, we generalize the algorithm proposed in [5] as the algorithm beta-VMA, parameterizing with beta the proposed threshold on the load allocated in a physical machine, and studying the effect of such parameter. 

In addition, we propose a new assignment algorithm, theta-VMA, that takes into account the rate at which a physical machine works. These algorithms are compared with the algorithm proposed in [5] and multiple state-of-the-art algorithms in different meaningful scenarios. Both beta-VMA and theta-VMA prove themselves as valid assignment algorithms, since they outperform the other algorithms in most of the cases.