Virtual Machine Consolidation for Performance and Energy Efficient Cloud Data Center Using Reinforcement Learning.
N.R.Rajalakshmi1, G.Arulkumaran2, J.Santhosh3
1N.R.Rajalakshmi, Associate Professor, Department of Computer Science and Engineering Computing, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science & Technology, Avadi, Chennai (Tamil Nadu), India.
2G.Arulkumaran, Assistant Professor, Department of Computer Science and Engineering Computing, Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science & Technology, Avadi, Chennai (Tamil Nadu), India.
3J.Santhosh, Assistant Professor, Department of Computer Science and Engineering Computing, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science & Technology, Avadi, Chennai (Tamil Nadu), India.
Manuscript received on 29 May 2019 | Revised Manuscript received on 11 June 2019 | Manuscript Published on 22 June 2019 | PP: 779-784 | Volume-8 Issue-3S, February 2019 | Retrieval Number: C11640283S19/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Cloud adoption and migration is growing everywhere because of its key benefits for cloud consumers and providers on economic, environmental and technological aspects. The augmentation of cloud computing service increases the size of the data center around the globe containing thousand of computing nodes. This increased large scale cloud data center has enormous quantity of electrical energy consumption which leads to huge operating cost and CO2 emission to the environment. The problem of high energy consumption and its effects are addressed here by presenting dynamic virtual machine consolidation using reinforcement learning. This reinforcement learning agent learns the knowledge from past history to attain the optimal policy. Hence, the number of active host is reduced by determining the host power mode based on current requirement. Hereby, the proposed work is compared with the competitive algorithms of Inter quartile Range maximum correlation policy (IQRMC) and Inter quartile Range Random selection policy (IQRRS). Experimental results show that, the required performance and energy level of data center is attained through proposed reinforcement learning.
Keywords: Machine Learning Cloud Performance.
Scope of the Article: Machine Learning