Research On A Cloud Task Scheduling Strategy Based On A Novel Sleep Mechanism

With the advent of an energy crisis and the rise in greenhouse gas emissions,how to achieve green cloud computing has attracted significant attention from academia and industry.In order to improve the energy conservation level of cloud data centers(CDC),decrease the operating cost of cloud service provider,and satisfy the response performance of cloud users,this paper studies the cloud task scheduling strategy and system performance based on the new sleep mechanism.Firstly,for the cloud data centers with high user service quality requirements,a sleep delay mechanism is introduced,and a cloud task scheduling strategy SD-SMSM-CTSS based on synchronous multiple sleep mode with sleep-delay is proposed(Cloud Task Scheduling Strategy based on Synchronous Multiple Sleep Mode with Sleep-Delay).By establishing the multi-server synchronization multi-vacation queueing model with vacationdelay,constructing the two-dimensional Markov chain,the steady state analysis of the system is carried out,and the performance measures such as the average response time of the cloud task and the energy conservation level of cloud system are given.Secondly,for the cloud data center with high system energy conservation level,a cloud task scheduling strategy SD-WT-SMSM-CTSS based on synchronous multiple sleep mode with sleep-delay and waking-up threshold is proposed(Cloud Task Scheduling Strategy based on Synchronous Multiple Sleep Mode with Sleep-Delay and Waking-up Threshold).Based on the shallow sleep situation of a physical machine,a multi-server synchronous N policy multi-vacation queueing model with vacation-delay is established.By using the matrix geometric solution methods,the energy conservation level and response performance of the system are quantitatively evaluated.Thirdly,based on the SD-WT-SMSM-CTSS,for the deep sleep situation of a physical machine,considering the waking-up process of a physical machine,a multi-server synchronous N policy multi-vacation queueing model with vacation-delay and set-up process is constructed.By establishing a balance equation,the steady-state distribution of the model is derived,and the system performance evaluation of SD-WT-SMSM-CTSS is carried out.Finally,the effectiveness of SD-SMSM-CTSS and SD-WT-SMSM-CTSS is verified by numerical experiments and simulation experiments.From the perspective of economics,compromise the energy conservation level of cloud system the average response performance of cloud users,construct the system utility function.For the SD-SMSM-CTSS,the Nash equilibrium behavior and social optimal behavior of cloud users are studied,and the social optimality is achieved by levying a reasonable admission fee.For the SD-WTSMSM-CTSS,based on the situation of shallow sleep and deep sleep of a physical machine,the system cost function is established,respectively.By improving the genetic algorithm,the optimization schemes of the strategy parameters such as the sleep parameter,the sleep delay parameter and the waking-up threshold are given.