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

With the rapid development of cloud computing and big data,massive data backup and migration have brought challenges to traditional data center networks.How to effectively schedule flows according to the internal traffic characteristics of the data center is the key technology to ensure the smooth operation of the data center.Compared with traditional networks,Software Defined Networking(SDN)is a new network architecture,which has the technical advantage of global network information visualization to adjust network management strategies.This provides new ideas for data center load balancing.Link load balancing problem for SDN-based date center in studied in this thesis.Firstly,in order to avoid the transmission collisions of elephant flows,a dynamic load balancing mechanism using particle swarm optimization(DLB-PSO)algorithm is proposed.Aiming to minimize maximum link utilization,we take advantages of global network topology information visualization of SDN controller,then get the global optimal solution by the particle swarm optimization algorithm.The simulation results show that,the proposed mechanism can effectively reduce flow delay jitter and improve network throughput.Secondly,and the classical routing algorithm-Global First Fit(GFF)which doesn't consider the remaining bandwidth of all the links will incure link fragmentation problem in traffic scheduling.In order to solve the above problem,a novel path selection algorithm based on the probabilistic is proposed.The algorithm makes it possible for the path with a small remaining bandwidth to be selected during the scheduling of elephant flows in the data center.The simulation results show that,the proposed algorithm can effectively reduce the average flow transmission delay and increase the average bandwidth utilization ratio.