Research On Network Load Balancing And Traffic Control Technology In SDN Environment

With the rapid development of computer network technology,the emergence of emerging technologies such as cloud computing and big data,the scale of data centers is constantly expanding,and network traffic is also exploding.The traditional IP network architecture cannot adapt to the rapid development of current SDN data centers.How to achieve efficient and flexible scheduling of traffic in the network and improve the overall performance of the entire network architecture has become a hot topic for scholars.The emergence of Software Defined Network(SDN)provides new ideas and methods for solving the problems existing in traditional networks.The SDN architecture separates the control plane from the data forwarding plane.The control plane maintains the view of the entire network and is responsible for formulating and delivering SDN network forwarding rules.The important function of the data forwarding plane is to implement data flow based on the flow table delivered by the control layer.Forwarding.Among them,the OpenFlow protocol is used for communication between the control layer and the forwarding layer.This protocol uses a flow table instead of the forwarding table in the traditional network.SDN reserves a programming interface for developers so that managers can manage SDN networks more simply and conveniently.At the same time,it can also meet the fine-grained demand of network users for network services.Aiming at the problems existing in the current data center,the corresponding research on the link load balancing and flow control algorithm in the traditional network is carried out.The main research contents and innovations of this article are as follows:(1)Based on Floodlight controller,a complete load balancing scheme is designed.The solution consists of the following functional units,which are information collection function units,traffic classification function units,scheduling decision function units,and routing delivery function units.The information collection function unit collects available bandwidth and bandwidth usage of each link in the SDN network.The traffic classification function unit classifies the traffic in the network according to the type and size by means of the Sflow network monitoring tool.Two improved algorithms proposed in this paper are integrated into the scheduling decision function unit to implement dynamic scheduling of traffic.The routing and sending function unit sends the calculated route to each OpenFlow switch in the SDN network in the form of a flow table.In this program,the OpenFlow protocol and the Sflow tool are used to combine four functional units to form a complete control loop.(2)For the traditional SDN load balancing algorithm with poor flexibility and low link utilization,this paper proposes an SDN link load balancing scheme based on improved particle swarm optimization for long data streams.The improved particle swarm algorithm(LB-PSO algorithm)is integrated into the Floodlight controller's scheduling decision function unit.The LB-PSO algorithm comprehensively considers the bandwidth between links and avoids the bottleneck link so that traffic is evenly distributed on each link.Finally,the LB-PSO algorithm was deployed on the simulation platform of Mininet+Floodlight.After simulation experiments verify that when the long data flow through the network,the algorithm can effectively improve the overall network throughput and link resource utilization.(3)For traditional flow control algorithms in a complex SDN network environment,it is easy to cause network congestion.This paper proposes an improved SDN flow control scheme based on improved Yen algorithm for short data flows.Link evaluation factors,normalized functions and link evaluation functions are introduced in the Yen algorithm.The algorithm increases the path diversity while considering multiple link assessment factors such as link delay,link bandwidth usage,link transmission speed,and switch hop count.Through experimental verification,when there are a large number of short data flows in the network,the K paths calculated by the Yen algorithm of the evaluation function not only increase the diversity of transmission links,but also reduce the transmission delay of short data flows in the network.