Research On Resource Scheduling Of Distributed Cloud Computing Based On Dynamic Stochastic Demand

In recent years,with the rapid development of cloud computing and distributed computing technology,various products of cloud computing have sprung up like mushrooms.Based on the technology of physical resource virtualization,research on resource scheduling at the cloud computing infrastructure and service level is the foundation for supporting various cloud computing products.However,current cloud computing resource management is still in its infancy,with many virtual machines,storage devices,network resources,and other infrastructure not being more effectively applied,increasing the burden of the global Internet,and generating huge server resource waste and energy consumption,which has become a major pain point restricting the development of cloud computing related technologies.Academia has proposed many innovative optimization methods for cloud computing resource scheduling,and the theoretical system for scheduling algorithms is gradually improving.Although there has been some development in resource scheduling algorithms at present,when regions and user numbers are large,the original mean-based resource scheduling algorithms still encounter problems such as load imbalance,resource waste,and inefficiency.Therefore,a distributed cloud computing resource scheduling algorithm based on dynamic random demand is studied,and the research contents are as follows:(1)Research scheduling methods under total resource constraints to provide appropriate input information for expansion methods.On the premise of meeting the quality-of-service requirements of cloud data centers,combined with the linear and nonlinear transformation effects of task scheduling methods such as request distribution in the demand aggregation process,to reduce fluctuations after demand aggregation.By combining various means in task scheduling,research on resource multiplexing methods that meet the quality-of-service requirements of cloud data centers,providing input information for subsequent scheduling methods based on total cost constraints and multiple system level scheduling methods in distributed cloud scenarios.(2)Research scheduling methods under total cost constraints,provide initial resource allocation schemes for scheduling at multiple system levels in distributed clouds,and use them to periodically re plan resource allocation schemes.Establish a common billing model and a common quality of service requirement model in cloud data center scenarios,and study automatic capacity expansion methods that are universal for random demands;Taking full account of the hierarchical characteristics of demand,we seek high-speed expansion algorithms to ensure availability in actual scenarios.(3)Research scheduling methods under multiple system layer architectures in distributed cloud scenarios,and quickly adjust regional resource allocation to meet dynamic random resource requirements based on total cost constraints.Establish a distributed queue model in the cloud data center scenario,consider the current system state,and combine the queue model to characterize the internal state of the system.Establish a distributed queue under quality of service and replica location constraints to obtain random demand information that the system needs to process.Combining multiple cloud resource rental methods such as on-demand instances and expected instances in cloud computing,aiming at unexpected demand bursts and fluctuating resource prices,considering the unsynchronized demand characteristics of various regions,and combining short-term and long-term demand change trends,further optimizing resource allocation.