文章摘要
范宝芝,王开宇,白小军,金顺福.面向异构物理机的云任务调度策略及性能优化[J].高技术通讯(中文),2021,31(10):1044~1054
面向异构物理机的云任务调度策略及性能优化
Task scheduling strategy in cloud computing and its performance optimization based on heterogeneous physical machines
  
DOI:10.3772/j.issn.1002-0470.2021.10.005
中文关键词: 云计算; 任务调度策略; 同步多重休假; 平均响应时间; 平均功率
英文关键词: cloud computing, task scheduling strategy, synchronous multiple vacation, average response time, average power
基金项目:
作者单位
范宝芝  
王开宇  
白小军  
金顺福  
摘要点击次数: 203
全文下载次数: 131
中文摘要:
      为了在保证云用户响应性能的前提下降低云系统的能源消耗,提出一种移动设备本地处理器持续工作、云端物理机内虚拟机同步休眠、不同物理机间虚拟机异步休眠的任务调度策略。针对云端异构物理机,建立多个同步多重休假的排队模型。利用拟生灭过程和矩阵几何解方法给出系统模型的稳态分布,导出任务平均响应时间和系统平均功率的表达式。实验结果表明,设置任务分配到移动设备本地的概率时,不同性能指标之间存在折衷关系。引入Logistic映射混沌机制改进传统的鲸鱼优化算法,给出最优任务分配策略,实现系统成本的最小化。
英文摘要:
      To reduce the energy consumption of the cloud system under the premise of ensuring the response performance of the cloud users, a task scheduling strategy is proposed. In the proposed strategy, the local processor of the mobile device continues to work, the virtual machines in a physical machine sleep synchronously, and the virtual machines in different physical machine sleep asynchronously. For the heterogeneous physical machines in cloud computing, a queueing model with a synchronous multiple vacation is established. By using the quasi birth-death process and the matrix geometric solution, the steady-state distribution of the queueing model is given, and the expressions of the average response time of the tasks and the average power of the system are derived. The experimental results show that there is a trade-off between different performance measures when setting the assigning probability of the tasks to the local processor. The traditional whale optimization algorithm is improved by introducing Logistic mapping chaos mechanism. The task scheduling strategy is optimized with the minimum system cost.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮