Research On Performance Optimization Of The Cloud Task Scheduling Strategy Based On Sleep Mechanism For Heterogeneous Physical Machines

With the increase of wireless network applications,the problems of limited storage and insufficient load capacity of mobile devices have become increasingly obvious.Cloud computing can provide support for task offloading of mobile devices.Taking into account the response performance requirements of cloud users and the energy consumption level of the cloud system,mobile devices can continue to work locally,asynchronous sleep between heterogeneous physical machines in the cloud,and virtual machines in the same physical machine sleep synchronously.Research on task scheduling strategies and performance based on sleep mechanisms optimization.First,for the cloud data centers with high system energy-saving requirements,a cloud task scheduling strategy PS-TSS(Periodic Sleep Mechanism based Task Scheduling Strategy,PS-TSS)based on a periodic sleep mechanism is proposed.Facing the remote cloud,establish multiple queuing models with multiple sleeps.For each physical machine,a two-dimensional Markov chain is constructed by combining the number of tasks with the sleep state and active state of the virtual machine,and the steady state distribution of the system is given.Secondly,for the cloud data centers with high user response performance requirements,a cloud task scheduling strategy OS-TSS(One-time Sleep Mechanism based Task Scheduling Strategy,OS-TSS)based on the one-time sleep mechanism is proposed.Facing the remote cloud,establish multiple queuing models with single sleep.For each physical machine,a two-dimensional Markov chain is constructed by combining the number of tasks with the sleep state and active state of the virtual machine,and the steady-state distribution of the system is given through the quasi birth-death process and the matrix geometric solution.Third,taking into account user response performance and system energy consumption level,a cloud task scheduling strategy FS-TSS(Finite-number Sleep Mechanism based Task Scheduling Strategy,FS-TSS)based on a limited sleep mechanism is proposed.For remote cloud,establish multiple adaptive multistage sleep queuing models.Combining the number of tasks with the sleep state,activation state and sleep times of the virtual machine,a threedimensional Markov chain is constructed,and the system steady-state distribution is given.Finally,using the steady-state distribution of the system,the performance index expressions such as the average task response time and the average operating power of the system are given,and the change trend of the average response time of the tasks and the average operating power of the system is revealed through system experiments.By setting the influencing factors of the average task response time and the average operating power of the system,the system cost function is constructed.The Logistic mapping chaos mechanism is introduced to improve the traditional whale optimization algorithm,and the optimal task allocation strategy is given to minimize the system cost function.