Research On Network Link Delay Measurement And Flow Table Management Method In SDN Data Center

In recent years,the development of cloud computing has caused the data center network traffic to show an explosive growth trend,requiring more flexible network management and network maintenance mechanisms.At the same time,the SDN data center network came into being.The highly centralized control capability of SDN simplifies the management and maintenance of data center networks,making data center networks more convenient and simpler in network measurement and congestion control.Therefore,this paper combines the advantages of SDN in network management and mainly studies the method of link delay measurement and flow table management in SDN data center network.Measuring network link delays requires placing measurement nodes and establishing probe paths in the network.The measurement node is responsible for injecting,receiving,and analyzing probe packets into the network,and the probe path is used to guide the probe packets along a pre-planned probe route.How to properly place measurement nodes and plan the detection path determines the detection cost and detection performance of the network link.In view of this situation,this paper conducts in-depth research on measurement node placement and path planning problems.By constructing the auxiliary bipartite graph,we transform the problem into the maximum flow minimum cost problem and design a heuristic algorithm based on the maximum flow.Based on the initial solution,the algorithm continuously tries to delete some measurement nodes to minimize the number of measurement nodes.In order to verify the performance of the algorithm,we compared the results of the heuristic algorithm with the optimal solution obtained by the linear programming.The results show that this algorithm can obtain the suboptimal solution or even the optimal solution in a short time.Moreover,the NS3 simulation results show that the round-trip delay of the link can be accurately measured by using the calculation results of the algorithm.In the flow table management method,because the data center network is easy to burst network congestion,in order to avoid the network getting into a worse situation or the response is expired,the network load balancing is required to respond quickly.In view of this situation,this paper conducts an in-depth study on load balancing of SDN data center networks under TCAM capacity constraints,and designs a heuristic load balancing scheme.This solution continuously reroutes the flows on the congested link through iteration to minimize the maximum link utilization.For this load balancing scheme,three re-routing algorithms with limited capacity of TCAM are designed.Among them,the shortest path-based fast rerouting algorithm first calculates a shortest path that does not pass through the congested link,and if the requirement is satisfied,the shortest path is used as the calculation result;the greedy routing algorithm based on K-shortest-paths first calculate K equivalent shortest paths that does not pass through the congested link,and then evaluated one by one,and the best one is used as the calculation result.The greedy routing algorithm based on the subsection route obtains the K equivalent shortest paths that does not pass through the congested link through the subsection route,and then evaluate one by one,and the best one is used as the calculation result.In order to verify the performance of the algorithms,we performed a simulation experiment on NS3.The results show that under the condition of limited TCAM capacity,the load balancing scheme can respond to network congestion quickly and effectively.Moreover,we compare three rerouting algorithms from the three aspects of load balancing effect,TCAM requirement and response time.