负载均衡的主导资源公平分配算法

随着使用云计算并行且可靠地处理计算问题成为一种趋势,各种云计算平台应运而生,在这些平台中,保证多种资源调度策略的公平性非常重要。主导资源公平分配算法DRF有效地实现了多种资源环境中的公平分配,但在资源分配过程中容易出现集群负载不均的情况。因此,提出在使用DRF算法分配资源过程中,通过集群中各节点的资源利用率情况对节点进行K-means聚类分析,根据聚类结果将资源分配给任务来提高集群负载均衡的能力。基于CloudSim 4.0实现了改进DRF算法的仿真实验,实验结果表明,负载均衡的DRF算法比原始的DRF算法以及基于层次分析法(AHP)改进的DRF算法更能有效地改善集群整体的负载均衡。

A fair load-balanced dominant resource allocation algorithm

With the trend of using cloud computing to handle computation problems concurrently and reliably, various cloud computing platforms have emerged. In cloud computing, the fairness of multiple resource scheduling strategies is very important. Dominant resource fairness (DRF) can effectively implement the fairness in multiple resource environments, however, it can introduce uneven cluster load in the process of resource allocation. Therefore, we propose a load-balanced dominant resource fairness allocation algorithm. It uses the DRF algorithm to allocate resources, perform K-means clustering on the slaves according to the resource utilization of each slave in the cluster, and allocate the resources to the tasks according to the clustering results to improve the load balance of the cluster. Simulations of the improved DRF algorithm are implemented on CloudSim4.0. Experimental results show that the proposed load-balanced DRF algorithm can improve the load balancing of the cluster more effective than the original DRF algorithm and the improved DRF algorithm based on analytic hierarchy process (AHP).