Research On Key Technologies Of Resource Allocation In Cloud Computing System

Internet information technology is flourishing,and the information stored on the Internet is increasing day by day.In the era of big data,science,finance and other fields are developing rapidly.People need to deal with data that the computer architecture could not do in the past.Cloud computing,as a technology that broke the traditional computer architecture and was born in order to adapt to the needs of today's labor productivity,its initial development started from a computer cluster.It achieved complex computing by gathering a large number of cheap servers and combining the management methods of distributed systems.The required computing power.Cloud service providers try to provide cloud computing as a commodity to customers.When the user's configuration does not meet the requirements,he can pay a certain fee to obtain the corresponding service through the cloud service provider to meet his own needs.In the industrial world,well-known companies such as Google are actively promoting the development of cloud computing.The key technology of resource allocation in the cloud computing system is a hot issue in cloud computing.The quality of its technology is directly related to the user experience when using cloud computing services and the service efficiency of cloud computing itself,which further affects the cloud service provider's profit.A good resource allocation algorithm can greatly increase the service efficiency of the entire cloud computing and minimize the energy loss of the data center.The quality of the resource allocation algorithm directly affects the quality of service(Qo S)and the violation of the SLA agreement.Various optimizations of cloud computing through the study of resource allocation algorithms have become a research hotspot in cloud computing.The essence of resource allocation is to allocate resources such as CPU memory of physical machines to virtual machines according to certain strategies.Therefore,most research on resource allocation integrates resources into virtual machines,and then considers the placement of virtual machines and physical machines.The paper comprehensively considers the energy consumption and performance loss of virtual machine placement,and converts the static placement of virtual machines into a continuous decision-making problem.Because the memory and other constraints need to be considered when virtual machine placement,the paper proposes two improved virtual machine placement algorithms based on reinforcement learning,RLRA algorithm and MCTSRA algorithm,based on memory constraints and other conditions.The formerimproves the reward value in reinforcement learning based on the established model,and sets the reward value to a value weighted by the performance loss number and energy consumption.The latter no longer uses the Monte Carlo search tree strategy selection method,but proposes a new selection algorithm based on the characteristics of cloud computing resource allocation.It is verified by comparison experiments with other algorithms that it has greatly improved in terms of energy saving.