计算机与现代化

• 数据库与数据挖掘 • 上一篇    下一篇

基于拍卖的数据中心资源匹配算法

  

  1. (1.中国科学院声学研究所国家网络新媒体工程技术研究中心,北京100190;2.中国科学院大学,北京100049)
  • 收稿日期:2018-04-09 出版日期:2018-10-26 发布日期:2018-10-26
  • 作者简介:王旭(1989-),男,山东龙口人,中国科学院声学研究所国家网络新媒体工程技术研究中心、中国科学院大学博士研究生,研究方向:云计算,边缘计算; 倪宏(1964-),男,研究员,硕士,研究方向:网络新媒体技术,宽带多媒体通信; 韩锐(1983-),男,副研究员,博士,研究方向:云计算。
  • 基金资助:
    中国科学院声学研究所青年英才项目(QNYC201715)

Auction-based Resource Match Algorithm in Data Centers

  1. (1. National Network New Media Engineering Research Center, Institute of Acoustics, Chinese Academy of Sciences, Beijing 100190, China;
    2. University of Chinese Academy of Sciences, Beijing 100049, China)
  • Received:2018-04-09 Online:2018-10-26 Published:2018-10-26

摘要: 云平台数据中心主机与负载均具有异构性,导致任务负载无法均衡利用主机各项资源。主机资源的非均衡利用最终造成总体资源利用率低,主机资源浪费,提高运营成本。针对云平台数据中心任务分配中各项资源无法均衡利用的问题,提出一种基于连续双向拍卖的虚拟机分配与迁移算法。该算法一方面利用多种启发式策略对数据中心主机和虚拟机进行筛选,将过载主机与欠载主机放入数据中心拍卖中;另一方面,构建买卖双方定价策略以及交易策略,形成完整的拍卖流程。同时,为解决多资源情况下的交易问题,提出基于资源匹配度的交易策略。仿真实验表明,文中方法通过引入资源匹配度,能够有效地匹配数据中心主机与虚拟机的各项资源,平衡各类资源利用率,提高整体资源利用率。

关键词: 云计算, 任务分配, 拍卖算法, 资源匹配, 虚拟机

Abstract: Hosts and workload in data-centers are heterogeneous, resulting in imbalanced utilization of host resources. This imbalanced utilization then results in poor overall utilization, waste of host resources and high operating cost. Aiming at the task assignment problem of data-centers in cloud platform, where different resources are not utilized evenly, a virtual machine allocation and migration algorithm based on continuous double auction is proposed. On the one hand, the algorithm uses a variety of heuristics to filter hosts and virtual machines, placing overloaded and underloaded hosts in data-center auction market. On the other hand, by rationally constructing pricing strategies and auction trading strategies, a complete auction process is formed. Furthermore, in order to solve the problem of multi-resource transactions, a trading strategy of auction process based on resource matching is proposed. Simulation experiments show that the proposed method can effectively match the resources of the data center host and the virtual machine. By introducing the resource matching, the proposed algorithm can balance the utilization of each kind of resource and improve the overall resource utilization.

Key words: cloud computing, task allocation, auction algorithm, resource matching, virtual machin

中图分类号: