Research On Observable Equilibrium Access Threshold Of Cloud Users With Sleep Mode

Cloud computing effectively aggregates virtualized computing resources based on the Internet,and uses aggregated data to provide various information services to users with different needs.It has the characteristics of dynamic,high performance-cost ratio and elastic scale expansion.But it is also associated with extremely high energy consumption.In order to avoid resource and energy waste,imbalance of service interests between cloud providers and cloud users which are caused by improper scheduling of virtual machines in cloud data center,the optimal access threshold strategy for cloud users in the cloud system sleep mode is studied based on different network businesses under the condition that information is fully observable.Firstly,aiming at the non-real-time network applications and the cloud systems with high energy saving requirements,this paper proposes a access threshold strategy of cloud user based on the periodic sleep mechanism when the information is fully observable.By solving the expected sojourn time of newly arriving cloud users in the dormant state and the awake state,a personal income function based on a linear revenue and expenditure structure is established,and the cloud user’s personal optimal access threshold is obtained.Different limited cache capacities are set in the wake-up state and the sleep state,then a multi-service desk synchronization and multiple vacation queuing model is built,and the steady-state solution of the system model is given to obtain the cloud user social optimal access threshold.Secondly,for cloud systems with high response performance requirements and low energy-saving requirements,the access threshold strategy for cloud user based on the single sleep mechanism is introduced within the cluster of the cloud system.In the situation where the information is fully observable,the cloud user’s personal optimal access threshold strategy is given by calculating the cloud user’s expected personal income.According to the different states of virtual machines,the limited access capacity of cloud users is set separately,and a multi-service desk synchronous single vacation queuing model is established.With the goal of maximizing the social benefits of the cloud system,the social optimal access threshold strategy of cloud users is given.Thirdly,considering the energy consumption and response performance of the cloud system in a compromise,the optimal access threshold strategy for cloud user based on the multi-level sleep mechanism is studied.In the situation where the cloud system information is fully observable,the equilibrium behavior of the system is analyzed,and the personal optimal access threshold strategy for cloud users is given.By establishing a multi-service desk synchronous multi-level adaptive vacation queuing model based on limited capacity,the steady-state solution of the queuing model is obtained,and the social optimal access threshold strategy for cloud users is given.Finally,according to the different sleep modes,we conduct system experiments based on theoretical derivation and analysis results by Matlab.For a given arrival rate and service rate,the results reveals the impact of system sleep parameters on the access threshold strategy when the virtual machine is in different states,and we study the relationship between the cloud user’s personal optimal access threshold strategy and the social optimal access threshold strategy.It also provides a theoretical basis for user access management of cloud systems in this paper.