【摘要】 在多资源的云计算场景中，用户提出的请求常涉及多种类型的资源。为实现集群负载均衡、系统资源利用与系统工作效率的提高，提出了RABBKC算法。根据内存资源与处理器资源的使用率统计结果，对云计算集群节点进行二分K-means聚类并找出各簇的聚类中心，将其中资源占用率最小的节点所代表的资源推荐至待分配的用户请求。再根据该节点资源总量算出能够分配的最大用户请求数目，结合主导资源公平的分配算法进行资源调度。仿真结果表明，算法可以有效实现集群中各节点的负载均衡，提高集群整体的稳定性，不仅有利于任务执行效率的提高，且对用户的响应更加公平，提供了良好的用户使用体验。更多还原

【Abstract】 In multi-resource cloud computing scenarios, user requests often involve multiple types of resources. In order to achieve cluster load balancing, utilization of system resources, and improvement of system work efficiency, the RABBKC algorithm was proposed. According to the statistical results of the utilization of memory resources and processor resources, the cloud computing cluster nodes were binary K-means clustered, the cluster centers of each cluster were found, and the resources represented by the node with the lowest resource occupancy were recommended to the user requests to be allocated. Then the maximum number of user requests that can be allocated was calculated according to the total resources of the node, and then combined with the dominant resource fair allocation algorithm for resource scheduling. The simulation results show that the algorithm can effectively balance the load of each node in the cluster, improve the overall stability of the cluster, improve the task execution efficiency, respond more fairly to users, and provid a good user experience.更多还原