Research On Task Scheduling And Resource Scheduling Algorithms In Cloud Computing

With the rapid growth of the Internet,there are more and more requirements for dealing with large-scale and vast amounts of data in the field of science research,engineering practice and financial industry.The demand for the computing power is far beyond the capacity of traditional computer architecture in these fields.Cloud computing is originally developed on the basis of the computer clusters.It reaches complex computing power by gathering a large number of cheap computers to form computer clusters and combining with distributed system software to carry out management.This way is different from the supercomputer.Cloud computing system provides the computing resources as a kind of commodity or service to the users,the users only need to pay the corresponding cost according to Pay-As-You-Go style.In industry,there are many well-known Internet companies are actively promoting the research and upgrade of cloud computing technology and development of related products at home and abroad.In academia,many scientific research units have established cloud computing research institutions.As main problem of upper layer of cloud computing,task scheduling problem need to be firstly addressed.The cloud computing task scheduling strategy which is good or bad can directly affect the user's degree of satisfaction and efficiency of cloud computing system.Good task scheduling method can effectively reduce the task completion time.It can also improve utilization rate of various computing resources,dramatically reduce energy consumption and save operating costs.On the other hand,the problem of energy consumption and the concept of green cloud provide new opportunities and challenges for the research of cloud computing.The resource scheduling algorithm which is good or bad will directly affect energy consumption and SLA(Service Level Agreement)violation rate.This dissertation will revolve this two core issues in the field of cloud computing: task scheduling problem and resource scheduling problem.In view of the task scheduling problem,the third chapter of this paper puts forward a kind of hybrid swarm intelligence optimization algorithm.The hybrid swarm intelligent optimization algorithm is an improvement of ant colony optimization algorithm verified effectively.By introducing users' spending index,the hybrid swarm intelligence optimization algorithm can cut down users' spending and keep higher efficiency.The hybrid swarm intelligence optimization algorithm provides a new selecting of task scheduling algorithm for private and community cloud service provider.In view of the resource scheduling problem,the fourth chapter of this paper has done work in three parts.The first part is that an improved vm placement algorithm is put forward in view of the vm placement subproblem.The second part is an improved vm selecting algorithm is proposed in view of the vm selecting subproblem,by using an existing vm selecting algorithm based on the bayesian model.The third part is to combine this two improved algorithms to get an comprehensive improved resource scheduling algorithm and to valid the resource scheduling algorithm.The proposed improved problem can save energy consumption and reduce SLA violation rate.At last,many experiments are performed on Cloud Sim which is a framework for modeling and simulation of cloud computing in view of each algorithm proposed above.In order to ensure the consistency of input data,the experiment design of the fourth chapter is based on data set build-in Cloud Sim.According to the experiment result,all algorithms proposed in this paper have done some improvement in view of the existing research results on corresponding common required evaluation index.