Research On Packet Forwarding Performance Model Of OpenFlow Network Based On Queuing Theory

Traditional networks integrate control logic and data forwarding functions into independent and closed equipment entities.It is difficult to deploy new technologies and protocols directly,which hinders the innovation and evolution of the network and meets the development needs of the future internet.As an innovative network architecture,Software-Defined Networking(SDN)has been widely applied in recent years,which decouples control logic from data forwarding,and flexibly defines packet forwarding behaviors in data plane through OpenFlow protocol.As the data transmission range becomes wider and the network traffic gradually expands,there are strict requirements for packet forwarding performance.Therefore,it is necessary to quantitatively evaluate the packet forwarding performance characteristics of SDN network,and design a feasible deployment strategy before practical application.To this end,based on queueing theory,this paper conducts modeling research on OpenFlow network performance,and formulates a controller deployment scheme to provide a reference for the application of SDN.For software defined Wide-Area Networks,a packet forwarding performance queuing model is proposed.Firstly,the typical deployment scenarios of SD-WAN are introduced,and analyze the processing diversity of different types of packets in data plane.Then we build an M/H3/1 queueing model for the packet processing of OpenFlow switches and deduce the expressions of its key performance metrics.Further,we establish the queueing model of OpenFlow-based packet forwarding in SD-WAN in combination with that of packet-in messages in controller clusters,and derive the closed-form solution of its average packet forwarding delay.Finally,experimental evaluation in virtue of simulation tools indicates that our proposed model can more accurately estimate the average sojourn time of different types of packets in comparison with existing models.Meanwhile,we compare the influence of different factors on average packet forwarding delay by numerical analysis,which provides guideline for actual SD-WAN deployments.Aiming at the controller placement problem,a green and efficient SDN controller cluster placement model is proposed.Firstly,the processing process of Packet-in messages by the controller cluster is modeled as an M/M/n queueing model to solve the average sojourn delay of Packet-in messages.Combined with the power consumption of controller cluster,an optimization model is constructed to solve the optimal number of controller placement.On this basis,the binary K-Center algorithm is used to realize clustering division,so as to select the placement location of the controller and construct the control domain.Then,when the controller is overloaded,a migration algorithm is designed to adjust the distribution of switches in the overloaded control domain,so that the controller cluster can achieve load balancing and ensure packet forwarding performance.Finally,the network topology is used for experimental evaluation.The results show that compared with the existing algorithm,the maximum propagation delay and average propagation delay of the proposed algorithm are slightly smaller than the existing algorithm,and better clustering effect is achieved,which provides a reference for the controller placement problem of software-defined network.