Research And Optimization Of Load Balancing Scheduling Algorithm Based On SDN Network

The deployment and application of large-traffic services such as big data and cloud computing have caused the data center traffic to increase dramatically.And the existing distributed network data center cannot provide flexible and convenient network load balancing functions.As a result,the problem of network data service transmission is affected by the continuous occurrence of network link congestion.The birth of Software Defined Network(SDN)has brought new breakthroughs to solve this problem.The software architecture of decoupling data layer and control layer can realize flexible network management and control,and its advantages of centralized control and external programmability meet the constantly updated network requirements.The controller makes use of its own network topology view to facilitate the formulation of network load balancing strategy.This paper focuses on learning and research SDN load balancing technology and does the following:I have conducted deeply research on how to reduce the data link layer transmission delay and packet loss rate and improve network throughput in SDN framework,and combining with the dynamic adjustable characteristics of traffic in data center network,the load balancing routing algorithm for stream scheduling is optimized.The algorithm calculates the remaining bandwidth and link congestion of all links according to the global network basic information monitored by the controller.When the load of the routing link in the network exceeds the link congestion degree,the optimized elite ant colony load balancing routing algorithm of the subject is invoked to recalculate the routing path for the newly arrived data stream,allocate the flow table resource,and ensure the link layer data flows are efficiently and quickly matched and forwarded.Because of the switch flow table resource capacity is limited,resulting in the uplink control link capacity between the data layer and the controller being easily congested due to frequent request messages,this paper adopts the flow redirection method to realize the purpose of balancing the Packet-In request message of the uplink control link by redirecting the flow of the newly arrived source switch and then uploading the controller to request the flow table resource.and constrains the redirection flow,optimizes and designs the MLFR algorithm.Finally,the virtual network experiment platform was designed and built by Mininet simulation platform,and the experimental verification of the proposed algorithm was completed.The results show that in the Fat(4)and Fat(6)topologies,the experimental results show that compared with the Dijkstra algorithm and the DLB algorithm,the optimized elite ant colony load balancing algorithm can reduce the link transmission delay to 0.018s.and reduce the packet loss to 14%,and increase the network throughput is to 76Mbps.Verify that the algorithm can improve network performance and reduce congestion of data streams;Compared with the SM algorithm and the OPT-C algorithm,the MLFR algorithm reduces the uplink load by nearly 50%,reduces the Packet-In message loss rate by 30%,and makes the switch flow entry resource distribution more uniform.