Research On Resource Configuration Management Of Distributed Heterogene Ous Network Based On SDN

The popularity of science and technology such as cloud computing,big data and artificial intelligence in the field of Internet has made the net-work environment more complicated.Providing reliable,secure and fast communication service quality has become an urgent problem for the net-work.The emergence of software-defined network provides a new idea for solving the existing network problems.The SDN architecture separates the control plane from the data plane,making data control and forwarding functions independent,and provides centralized control and programmable functions,which greatly promotes the innovation of network technology.First of all,this paper studies a new generation of SDN network ar-chitecture and its key technologies,introduces the architecture knowledge of SDN,OpenFlow technology and Floodlight controller related principles,and then introduces and analyzes in detail the traditional network load bal-ancing algorithm and dynamic Routing algorithm.In view of the limita-tions of the traditional algorithms,this paperpresents a solution combining load balancing and link routing based on SDN technology.The new pro-gram aims to solve the problem of network resource allocation and im-prove the utilization of network resources.Aiming at the structure of con-troller cluster in SDN network,this paper presents two kinds of load bal-ancing architecture-distributed load balancing and centralized load bal-ancing,and analyzes the advantages and disadvantages.Aiming at the load balancing algorithm,this paper proposes a load balancing algorithm based on predictive allocation scheduling.According to the actual network re-quirements,the algorithm sets the overload threshold of the controller and makes an overload prediction considering the number of tasks to be pro-cessed,the Packet-in packet rate and the processing task rate.If the pre-dicted result is greater than the overload threshold of the load,the control-ler begins to collect the load information of the other controllers in the cluster,calculates the initial load value,and updates the load information table to prepare the data for the controller load balancing calculation,so as to ensure the load Balanced efficiency.Secondly,a new routing scheme is proposed for the traffic routing with short lifetime and small amount of data in the network traffic center.The existing updating strategy mostly updates the link information when the SDN periodically acquires the global network parameters and can not reflect the real link resource situation in real time.Path calculation uses a typical shortest path routing algorithm-Dijstra algorithm.Combined with SDN to obtain the global network topology characteristics,the algorithm is strictly limited.In the process of path finding,the link bandwidth re-sources are updated in time according to the optimal path delay and the lifetime of the data flow.Finally,this paper uses Mininet network simulation platform to setup SDN experimental environment.First,a centralized and distributed system structure is constructed,and the load balancing algorithm is simulated re-spectively.The simulation results show that the performance of the distrib-uted scheduling algorithm based on prediction is greatly improved com-pared with the dynamic adaptive load balancing algorithm(DALB),and the final load effect is better.Secondly,aiming at the fast update dynamic routing strategy,this paper uses Mininet simulation platform to create a custom network topology,experiments on the improved dynamic routing strategy,and compares the service rate and bandwidth utilization of differ-ent strategies.The simulation results show that the dynamic update of dy-namic routing strategy can improve the link bandwidth utilization and user service quality.