Research On Key Technologies Of High Availability Control Plane In Software-Defined Networks

Software Defined Networks(SDN)can effectively enhance the flexibility and convenience of the network by separating the control plane and the data plane.However,there are many problems with the popularity of SDN.Designing and implementing high-availability softwaredefined networks has become a research focus.As the core component,the control plane is the prerequisite for the high-availability software-defined networks.Many researches have been proposed to enhance the high-availability of the control plane from the perspective of control efficiency.The technology,such as high-performance controllers and high-performance control clusters,can optimize the control plane to a certain extent through resource optimization and elastic control mechanism.Recently,as the cyber security deteriorates,designing a secure and reliable control plane is an urgent need for the high-availability control plane.To achieve this goal,the fault-tolerant control plane is proposed.The fault-tolerant technology can improve the robustness of the control plane through the redundant,dynamic and heterogeneous architecture.However,there are many challenges while applying the fault tolerance technology to the SDN mainly including two aspects:(1)In terms of networking,how to deploy control resources in the fault-tolerant control plane,how many control resources are deployed,and how to design reasonable and effective failover mechanism are important issues to improve the availability of fault-tolerant control plane;(2)In terms of architecture,how to use the new technologies and new architectures to improve the performance of the fault-tolerant architectures is also a problem.In view of the above problems,this dissertation studies the key technologies of the highly availability control plane in the software-defined network.In terms of networking,the dissertation uses the resource optimization and the failover mechanism to manage the control resources in SDN,which enhances the high-availability of the control plane.In terms of architectural,the dissertation is inspired by the moving target defense technology and the mimic technology.Through the dynamically scheduling of the heterogeneous variants,the performance of the fault-tolerant control plane can be improved.The main contribution of this dissertation are as follows:1.Focused on the high deployment cost brought by the fault-tolerant control plane,we take switches' fault tolerance requirement,communication delay between switches and controllers,communication delay between controllers,and controller load capacity into consideration.The minimum fault-tolerant coverage model is proposed.Then a controller placement algorithm based on the local search algorithm is designed.This algorithm can get the optimal controller placement by continuously generating neighbor controller placement.The simulation under different scale networks shows that the algorithm can deploy a control plane with the low deployment cost while meeting the fault tolerance requirements of the control plane.2.For the lack of a reasonable failover mechanism for the fault-tolerant control plane in a multi-controller failure scenario,we build a master-backup controller mapping model based on the fault controller sequence.The load balance of the remaining controllers and the delay between the controller and the switch is estimated when multiple controllers fail.To avoid isolated switch,a heuristic algorithm based on the label propagation is designed.The algorithm can update and optimize the fault controller sequences to obtain the optimal placement.The simulation shows that the proposed algorithm can not only alleviate the deterioration of the delay and load balance,but also that the switch will not become the isolate node.3.For the lack of weight of the backup control path,we estimate the impact of control path,and then quantify the failover cost when control path failure occurs.Then,an integer programming model is proposed by combining the deterioration delay and the failover cost.In order to solve the model,an optimization algorithm based on simulated annealing algorithm is designed.The simulation shows that compared with other algorithms,the proposed algorithm can effectively reduce the network's failover cost and enhance the resilience of the fault-tolerant control plane.4.Addressing at the problem of homogeneous vulnerabilities in heterogeneous variants,a homogeneous-vulnerabilities-aware dynamic fault-tolerant control plane is designed.While maintaining the state of the heterogeneous controllers,the dynamic fault-tolerant control plane can also configure and schedule the fault-tolerant control plane.First,in order to configure a reliable heterogeneous control plane,the method is designed based on the similarity and uses the number of probes to model the risk of fault-tolerant control planes.The simulation shows that the proposed algorithm can quickly select a suitable combination of controllers from many heterogeneous controllers to reduce the risk of the control plane.Then in order to ensure the availability of the heterogeneous fault-tolerant control plane in scheduling process,a method is proposed to analyze the gain-cost ratio of the scheduling process.After constructing an integer programming model,we design the algorithm to solve the problem.Compared with other algorithms,the simulation shows that the proposed algorithm can reduce the homogeneous vulnerability risk of the faulttolerant control plane with a lower scheduling cost.Finally,a prototype system proves the effectiveness of the proposed dynamic heterogeneous fault-tolerant control plane.