Research And Implementation Of Load Balancing Subsystem Based On SDN Data Center Network

Chinese technologies such as cloud computing,big data have developed at top speed in recent years,which led to the rapid growth of traffic in the data center network.Facing the enormous network data traffic,reasonably dispatch the traffic in the network to achieve the load balancing of network resources is one of the current research hotspots.Because the traditional data center network is unable to obtain the state of the current network,it usually adopts a static scheduling strategy.It is easy to make multiple elephant flow into the same path to cause the network load to be uneven.Essence Software Defined Networking(SDN)can separate the control functions of traditional network devices into controller,and can obtain the network status through the controller,so that the flow scheduling is dynamically scheduled according to the current state,which can solve the above problems well.Therefore,this article mainly uses SDN technology to solve problems in traditional networks.This article first proposes a dynamic flow scheduling algorithm based on high response ratio.For the problems existing in the traditional static flow scheduling method,the algorithm will comprehensively consider traffic characteristics and the current network status to perform traffic scheduling.First of all,the algorithm will calculate the response ratio of the current flow according to the size of the data flow and the waiting time,and calculate the shortest path set from the source end to the destination end by selecting the flow with the largest response ratio each time,and then calculate the weight of each path by considering the current network delay,the Bottleneck link bandwidth of the path and the average remaining bandwidth of the path,so as to select the optimal path according to the weight,and realize the load balance of the network link.The simulation results show that the algorithm has a certain degree of improvement over the traditional algorithm.Because the traffic in the data center network has the characteristics of sudden and unstable,only through the initialized scheduling algorithm cannot avoid the occurrence of network congestion.In response to this problem,this article proposes a load balancing algorithm based on DDQN implementation.This algorithm will be regular To detect the current network status,when the load of the link exceed the threshold,it means that the current network’s load is uneven.Then screen out the elephant flow collection in the current link,use the elephant flow collection,the status of the current network,and the path of each stream to train the DDQN model.Finally,the trained model is used to select the optimal path for each flow.Simulation results show that the algorithm can reduce network congestion to a certain extent and effectively improve network throughput and link utilization.Finally,based on the above algorithm model,this paper designs and implements a load balancing subsystem based on SDN data center network,which provides platform engineering support for the previous algorithm research and provides it to network managers through web applications.This paper introduces the overall design of the system,the design and implementation of each functional module,and the typical workflow,and deploys and tests the prototype system of the platform.The test results show that the system can provide a visual interface for network managers,so that they can clearly observe the load of each link in the network,so that network managers can make more reasonable use of network resources.