An energy-aware resource deployment algorithm for cloud data centers based on dynamic hybrid machine learning

Abstract

To meet the ever-increasing requirements of cloud users, cloud service providers have further increased the deployment of cloud data centers. Cloud users can freely choose the cloud data center that suits them according to their own business characteristics and budget expenditures. This requires cloud service providers to continuously improve service quality and reduce usage costs to expand their own user base. Mature cloud service providers will continuously optimize cloud tasks and virtual machine deployment methods to increase physical machine utilization and reduce cloud data center energy consumption. However, existing virtual machine deployment algorithms usually have low utilization of physical machines or high energy consumption of cloud data centers, thereby reducing the frequency of use by cloud users and the benefits of cloud service providers. This paper systematically analyzes virtual machine and physical machine models. At the same time, the K-means clustering algorithm for unsupervised learning and the KNN classification algorithm for supervised learning are expanded to establish a dynamic hybrid resource deployment rule. Then, an energy-aware resource deployment algorithm for cloud data centers based on dynamic hybrid machine learning (EHML) is proposed based on the theory of machine learning. This algorithm reduces energy consumption by increasing the average utilization of physical machines. Finally, the experimental test results show that the average utilization of physical machines and energy consumption of the algorithm are significantly better than those of the comparison algorithms.