Resource Allocation For Reliable Communication Between Controllers And Switches In SDN

Software Defined Networking(SDN) is a new architecture of network,which has attracted a lot of research and attention of industry and is regarded as an important technology of future network. SDN splits the computation modules and data forwarding modules of traditional network devices, and computation modules are deployed on a centralized controller,which is well supported by OpenFlow protocol. In SDN, switches must work according to the flow table entries. Thus, how to ensure the reliability of the communication between controllers and switches is a key problem in SDN.The reliability of communication between controllers and switches mainly depends on two kinds of factors, one is the topology of SDN network, the other is controller placement. So in the network design phase,when the topology of SDN network is determinate, reasonable controller placement can effectively improve the reliability of communication between controllers and switches. To ensure the reliability of communication further, we also need to prepare backup resource for controllers and control paths, in case of the failure of controllers and control paths. Especially, some failures are caused by disaster happened in the network environment. A disaster usually causes a regional failure of the network. In this condition, the backup resource may be influenced while the work resource is influenced too, so the failure cannot by recover by the backup resource.To improve the reliability of the control network in SDN, we need to allocate resource for the control network reasonably. In this paper, we discuss the controller placement problem and resource backup problem about resource allocation for the control network, the main research contents include:(1) A min-cover based controller placement approach is proposed,which is aimed to place several controllers appropriately in network design phase to ensure the reliability of control network, and the required delay between controllers and switches can be satisfied. The approach firstly defines the metrics to measure the reliability of control network, and propose the notions of neighborhood and min-cover, based on which we transform the controller placement problem into solving the min-cover of network, finally we use an algorithm based on binary search to solve the problem.(2) A resource backup approach against regional failure is proposed.We firstly formulate the regional failure to measure the influence. For the controller backup problem, we suppose the main controller as the event source, and we consider how regional failure influence the vertex in network and the delay performance of each vertex to choose the location of backup controller. For the control path backup approach, we suppose the event source is at one link of the control path, and we choose the backup control paths by considering the survivability of backup control paths.