摘要
基于预留机制的服务器动态开启/关闭(dynamic powering on/off servers,DPS)策略采用静态设置的任务请求数阈值,可能造成服务器状态频繁切换从而导致性能下降、能耗上升。对此,提出一种基于动态阈值的服务器唤醒策略。首先,用具有不耐烦任务的排队模型对云计算系统的任务调度进行建模,分析系统中的平均任务背叛数和能耗成本,提出任务请求数阈值动态调整策略;然后,根据服务器所在冷点区域和当前关闭时长选择服务器进行唤醒。仿真结果表明,与基于静态阈值的服务器唤醒策略相比,本文策略能够保证任务的平均响应时间,并有效降低云计算系统的能耗开销。
The dynamic powering on/off servers （DPS） strategy based on reservation sets the static threshold of the task request number beforehand, which may cause frequent switching of the server status. To solve this problem, a strategy of servers awakening based on the dynamic threshold in cloud computing is proposed. Firstly, the queuing model with impatient tasks is introduced to model the task scheduling in the cloud compu- ting system, and analyze the average number of task betrayal and the cost of power consumption, thereby the strategy of dynamicly adjusting the threshold of the task request number is presented. After that, a server is chosen to be awakened according to the cold area where the server located and the length of its shutdown time. The simulation results show that the proposed strategy can ensure the average response time for tasks and reduce the energy cost in cloud computing system efficiently.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2015年第6期1437-1445,共9页
Systems Engineering and Electronics
基金
江苏省科技支撑项目(BE2012849)
江苏省研究生科研创新计划(CXZZ12_0483)资助课题
关键词
云计算
服务器动态开启/关闭
动态阈值
排队论
不耐烦任务
cloud computing
dynamic powering on/off servers （DPS）
dynamic threshold
queuing theory
impatient tasks
