Power-aware Virtual Machines consolidation architecture based on CPU load scheduling

Abstract

One of the concepts most currently emerging, in the world of information technology is Cloud Computing. While it provides many advanced features, there are still some shortcomings, such as the exorbitant operational cost. It is indeed at the center of several fields of research. One of them is the energy, under its different forms. The management of energy consumption in the world of Cloud is the one on which we concentrated through this paper, that is part of a work around the energetic aspect of Cloud Computing. The field of Green Computing is becoming more and more important in a world of limited energy resources and a growing demand for greater computing power, it therefore represents one of the major concerns, that we are dealing with in this paper, in order to optimize the consumption of electrical energy in a virtualized Cloud Platform, while respecting the requirements of SLAs. We propose a framework which offers effective improvements in a scalable Cloud architecture. A solution is to dynamically consolidate the Virtual Machines (VMs in the paper) of the platform in a lower number of nodes (hosts), during the time of low load situations, in order to shutdown unsolicited nodes, in the condition of providing an acceptable level of availability and response time and storage capacity so as not to penalize the user and respect the constraints of the service contracts.