带优先权用户和半休眠机制的云系统节能分析

Energy Saving Analysis of Cloud System with Priority Users and Semi-Sleep Mechanism

摘要: 针对云用户呈指数增长现状，云环境下为了在保证用户服务质量的同时降低系统能耗、提高系统收益的问题。提出运用带抢占优先权和服务台同步工作休假的M/M/c排队模型对云系统性能和系统收益进行建模分析。建立三维Markov过程，得到其无穷小生成元矩阵，采用拟生灭过程理论求出系统的稳态分布，得出两类顾客的平均队长、内存溢出率、系统总能耗等稳态性能指标，并建立相应的系统收益函数。通过数值模拟探讨参数对系统稳态性能指标的影响，研究VM半休眠机制对系统总能耗的影响以及参数对系统收益函数的作用规律。在保证两类顾客服务质量的前提下，通过动态调整VM数、内存空间和服务率来降低系统总能耗、增加系统收益。

Abstract: In order to reduce system energy consumption and improve system profit while ensuring user ser-vice quality, this paper proposes to model and analyze the cloud system performance based on the M/M/c queue with preemptive priority and server synchronous vacation. The three-dimensional Markov process is established, its infinitesimal generator matrix is gained, the steady-state distri-bution of the system is gained by using the theory of quasi birth-and-death, the steady-state per-formance indicators such as the average queue length of the two types of customers, memory over-flow rate, and total energy consumption of the system are gained, and the corresponding system benefit function is established. Through numerical simulation, the effect of parameters on the steady-state performance index and profit of the system is discussed, and the effect of the semi-hi- bernation mechanism of virtual machine on the total energy consumption of the system are studied. Under the premise of ensuring the service quality of customer service of the two types, it is to re-duce the total energy consumption of the system and to improve system benefit by dynamically ad-justing the number of virtual machines, memory space and service rate.