Research On Energy Consumption Scheduling Algorithm Based On Cloud Computing Platform

With the rapid development of information technology,such as parallel computing,distributed computing and virtualization technology,the original IT architecture system has been gradually replaced,and cloud computing technology has risen rapidly.Cloud computing integrates the scattered system resources into a huge resource pool,and the resources of the data center are provided to the outside through resource sharing.Users can quickly obtain computing resource services by paying on demand,without considering the maintenance of the underlying infrastructure.Due to the increasing number of cloud computing users and the increasing demand for hardware resources,the scale of data centers is also expanding,and the energy consumption and resource waste generated by data centers are increasing.How to reasonably schedule resources in a complex cloud environment,improve the resource utilization rate of the cloud platform,and reduce the overall energy consumption of the data center has become the core problem that needs to be solved urgently in today's cloud computing research.To solve the above problems,this thesis focuses on the initial placement and migration of virtual machines based on Open Stack cloud platform by analyzing its native resource scheduling strategy and deeply studying virtualization,aiming at optimizing the overall power consumption of the platform and avoiding hot spots through scheduling algorithm optimization.The main work and innovation of this thesis are as follows:To solve the problem that the traditional virtual machine scheduling model has a single scheduling objective,the models with the optimization objectives of improving resource utilization,load balancing and reducing cloud platform energy consumption are constructed respectively,and the evaluation functions are designed according to these three models.To solve the multi-objective optimization problem of resource scheduling in cloud platform,a virtual machine placement strategy which combines genetic algorithm and particle swarm optimization algorithm is proposed.Combining the two algorithms in series,genetic algorithm has a wide search range and particle swarm optimization algorithm has the advantages of good local search ability and fast convergence speed,and a dynamic adaptive inertia factor is introduced to adjust the global search and local search of the algorithm according to the particle fitness,so as to quickly find the optimal solution of virtual machine-to-server mapping and find the optimal resource scheduling scheme.The experimental results show that compared with the relatively simple heuristic algorithms such as genetic algorithm and particle swarm optimization algorithm,the proposed algorithm has better resource utilization,more balanced load and lower energy consumption.For the problem of local hot spots in servers,a virtual machine migration algorithm based on random forest prediction is proposed.By predicting the future energy consumption changes of servers,the threshold of energy consumption is set,and the virtual machines in servers whose energy consumption exceeds the threshold are migrated,thus avoiding the occurrence of server hot spots.The experimental results show that the algorithm can effectively avoid the hot spots in the server.