Dynamic nearest neighbor resources classification algorithm for multiple cloud data center based on natural clustering rule

Abstract

Abstract Cloud service providers need to reduce the operating costs and energy consumption of cloud data center (CDC) by optimizing scheduling algorithms, and ultimately reduce the cost of cloud users. However, the existing scheduling algorithms are less effective in dealing with the scheduling problems of multi-cloud data center (MDC). This paper systematically analyzes the MDC model and physical machine (PM) utilization. Secondly, using the idea of K-means clustering algorithm in machine learning, natural clustering rules are proposed to complete automatic clustering of PMs. Then, the supervised learning KNN classification algorithm is extended and the dynamic KNN classification rules are established accordingly. Finally, a dynamic nearest neighbor resources classification algorithm for multiple CDC based on natural clustering rule (DNSC) is proposed. Comparing the algorithm with the comparison algorithm shows that the algorithm comprehensively considers the resource parameters of the MDC and ultimately reduces the energy consumption of the MDC.