Design Of Cross-Domain Experimental Cloud Platform System And Resource Allocation Strategy Research

With the rapid progress of information technology and terminal equipment,the standards and requirements for application services are becoming higher,and at the same time,the demand for high-performance computing services in the industry is increasing.Cloud computing has become the first choice for many applications due to its advantages of high computing power,low service cost,high scalability,accessibility and availability.In cloud computing environments,different users have access to heterogeneous resources with different characteristics that these resources are geographically in different areas.In addition,in the real cloud computing scenario,most enterprises deploy their services in a multi-cloud or multi-domain environment for disaster recovery purposes.Therefore,cross-domain resource allocation in cloud computing becomes a major challenge to achieve multi-domain collaboration in cloud computing.At the same time,the market size of cloud computing continues to expand,and its economic factors cannot be ignored.The efficient allocation of multiple resources and the economic benefits to all parties involved in the transaction need to be considered in the allocation of resources in the cloud computing market.In order to explore the effectiveness of resource allocation algorithm based on methods of economics in dealing with cross-domain multi-resource allocation in real system,this thesis designs an experiment cloud computing platform system deployed in multiple regions.On this basis,a cross-domain resource allocation strategy based on combinatorial double auction is proposed,which can increase overall utility and avoid malicious bidding.Finally,the system is deployed in a practical environment and the proposed resource allocation strategy is verified,which has practical significance.The main work of this thesis is as follows:(1)A virtual experiment cloud platform system is designed and implemented,which can centrally manage experimental network devices such as switches and virtual machines.It can establish experiment topology on graphical interfaces and deploy the topological nodes on physical devices as virtual machines,so that users can log in to virtual machines remotely to conduct their experiments.(2)In order to make the system have the characteristics of multi-location deployment and disaster recovery,based on the above experimental cloud platform system,a crossdomain service architecture based on centralized control mode is designed to realize the multi-regional deployment of the cloud platform and support inter-domain collaborative experiments.Its general framework can be described as follows: a master control cloud platform system is set up to control multiple sets of single-domain experiment cloud platform systems.The master control platform can integrate resources of each single-domain system,negotiate cross-domain resource request tasks initiated by a single domain,allocate free resources of other domains,and complete cross-domain resource allocation.(3)Based on the characteristics of multi-resource attributes,multi-domain participation of cloud platform and the background of the existing classical auction model,a strategy based on combinatorial double auction is proposed.The proposed strategy can help the master control platform complete cross-domain resource allocation,so as to optimize the system economic benefits and improve the resource utilization rate.In addition,a data transfer module is designed,whose function is to import the resource data which is waiting to be allocated on the cloud platform into the simulation platform for resource allocation according to the proposed algorithm,and then export the allocation results to the cloud platform.(4)The proposed resource allocation strategy is implemented in CloudSim cloud computing simulation platform and compared with other resource allocation algorithms.The results show that the proposed resource allocation strategy has advantages in user utility,resource allocation rate and incentive compatibility.Finally,the function of the data transfer module is verified.