云数据中心的多维资源调度策略探析

摘要

资源调度问题是数据提高运行效率的关键。然而目前大部分任务调度算法都聚焦于单一维度的资源，忽略了多维资源( CPU、Memory、Disk、Storage)同时消耗的实际情况。本文将此多维资源调度问题归纳为多维背包问题，并且提出了相应的MS调度算法。为了展示MS的优势，在WMware平台做了仿真实验，实验结果表明：MS算法能取得较高的服务器综合资源利用率。%Resource scheduling is crucial to data centers. However, most existing scheduling algorithms focus only on one-dimensional resource, ignoring the fact that multiple resources ( CPU, Memory, Storage and bandwidth) are consumed simultaneously. This paper maps such a resource scheduling problem to a multi-dimensional knapsack problem and pres-ents MS- a relevant scheduling algorithm. To demonstrate the advantage of MS, the paper has implemented MS on VM-ware platform and the experimental results show that MS algorithm can achieve higher integrative resources utilization per-centage of servers.