Research And Implementation Of Resource Scheduling Scheme Based On Cloud Computing

With the rapid development of cloud computing technology,the traditional IT architecture is gradually being replaced.Leveraging the virtualization technology,Cloud computing virtualizes many homogeneous or heterogeneous physical resources,construces a virtual resource pool and provie service to users.Through the Internet,Users can obtain virtual resources according to their own demands.At present,the number of cloud users and the demand of cloud users has increased dramatically,the scale of cloud data centers has also expanded unceasingly.Today,how to rational schedule physical resources in a dynamic heterogeneous cloud environment,improve the performance of the cloud platform and protect user needs is becoming one of the urgent problems in the cloud computing field.In view of the above problems,the thesis studies the virtualization technology and existing resource scheduling algorithms in cloud environment.In this paper,to load balance and improve the utilization rate of resources and reduce the energy,a further research is made on the virtual machine resource scheduling schemes based on the IaaS cloud platform OpenStack.The main work and innovation of this thesis are as follows:(1)If an improper resource scheduling scheme is designed,it will lead to waste of physical host resources,load imbalance between physical machines,and increased energy consumption of the cloud platform.This thesis built three models based on optimized targets of load balancing,improving resource utilization,and reducing the energy consumption of the cloud platform.And then this paper designed an evaluation function with three models.(2)A virtual machine scheduling scheme based on Hybrid Genetic Algorithm And Quantum Particle Swarm Optimization(HGQP)is proposed.Because genetic algorithm(GA)can search in a wide range and it has excellent global search capabilit,quantum particle swarm optimization algorithm(QPSO)has the advantage of fast convergence speed and excellent local search capability.Through the serial fusion of algorithms,divided the number of iterations into two parts,and the first half of the iteration uses GA.After reaching 1/2 of the number of iterations,the QPSO is used.This fusion algorithm can speed up the search for the optimal solution to the mapping from virtual machine to physical host,and finally the optimal resource scheduling scheme is obtained.(3)Aiming at the defect of QPSO in the resource scheduling,it is esay to fall into the local optimum and the convergence speed will slow down in the later iteration.This paper introduced an adaptive method into QPSO,and used the particle fitness value to adjust contraction-expansion coefficient which is the only alterable parameter in algorithm.The average optimal position of the algorithm is improved by introducing elite strategy and adding weight factor when calculating the average value.(4)A comparative experiment on the HGQP algorithm was conducted based on the cloud computing resource scheduling simulation environment Cloudsim.The experiments result show that the HGQP proposed in this paper has shorter task processing time,better CPU utilization and lower energy consumption,compared with the traditional single heuristic algorithms such as genetic algorithm,particle swarm algorithm and Least-Full-Fit algorithm(LFF),which verifies the effectiveness and rationality of the algorithm.(5)Based on an open source cloud computing management platform project Openstack realizes a cloud computing resource scheduling platform based on the HGQP resource scheduling scheme.Designed the resource management module,resource monitoring module and resource scheduling module.The functional test results show that this platform realizes the unified management,monitoring and deployment of the underlying virtual resources and physical resources,and can provide automated and integrated operation and maintenance methods for businesses and applications.