A Routing Optimization Considering Both Control And Data Traffic In SDN

The rapid growth of traffic has posed great chanllege to data center network.On the one hand,it has huge energy consumption and is bad for the environment,economy and sustainable development.On the other hand,it is not easy to achieve timely and reliable network services with restricted resource.Routing mechanism can optimize the forwarding services by analyzing,predicting and scheduling traffic.However,with the traditional distributed forwarding and control structure,networks cannot achieve fine-grained way of resource scheduling,which cannot meet the demands of network technology development and business application innovations.Therefore,we need a new network architecture to inject vitality into the Internet.Software defined networking(SDN)is an emerging networking paradigm that separates the network control plane from the data plane.The complex control function has been centralized to the controller so that the network switches only need to perform simple and efficient actions.Controller can have a whole view of the entire network,which can schedule the traffic in a more fine-grained way,reduce the complexity of routing design and increase the performance of routing mechanism.This paper mainly studies SDN routing mechanism according to two aspects,namely the data plane and control plane.First of all,we studied the energy-saving routing problem in data plane which employed SDN paradigm.A novel energy-aware routing approach was proposed,it focused on in-band SDN environment,where control traffic and data traffic share resources in the same network.The approach added constraint of control traffic in traditional energy-saving routing problem statement,and came to the optimization objective referring to the energy model of SDN network device.Secondly,after the realization of energy-saving routing,network topology and link load changes,which will affect the original control network.Therefore,we also studied the control network optimization based of load and modification cost.We defined the conception of criticality of link,based on which we computed k minimum spanning trees concurrently.By evaluating the k trees using both load and cost factors,we came to the optimal solution and maintain the reliability of control network.