|
چکیده
|
Cloud service providers provide efficient cloud services by continuously building new data centers. Container data center (CDC) is a fast-deployment data center model. Because its construction method is different from the traditional data center, its energy consumption optimization problem has become a core issue. This paper optimizes the performance of CDC by combining scheduling. First, the core point threshold calculation method of the density-based spatial clustering of application with noise (DBSCAN) algorithm is modified, and the dynamic core point threshold DBSCAN (DCTD) algorithm is proposed. The algorithm can complete the clustering of different racks in CDC. Secondly, the K value selection principle of the K-Nearest Neighbor (KNN) algorithm is modified, and an adaptive KNN (AKNN) algorithm is proposed to complete the classification of the virtual machine (VM) to be deployed. Thirdly, the prevent hotspot equalization mapping algorithm (PHE) is proposed according to the parameters of the VM and the rack, which can complete the selection of the rack and the mapping on the rack. Finally, through the above combined optimization method, a combined optimization scheduling mechanism of CDC based on clustering algorithm in machine learning (COCM) is proposed. In order to evaluate the effect of the algorithm, it is compared with other algorithms and verified. The results show that the algorithm has lower energy consumption and higher balanced utilization of physical machine (PM).
|