Research On Resource Management Strategies For Multi-type Services In Software Defined Networking

Software Defined Networking(SDN)is a new type of network paradigm used to manage the flow transmission of various networks.In SDN,the controller in the control plane can centrally manage flow transmission,and the switch in the data plane is only responsible for data forwarding.Therefore,the mapping relationship between the SDN controller and the switch will play an important role for the performance of the network.In recent years,with the rapid development of Internet of Things and 5G networks,more and more high-bandwidth and low-latency services have entered people's lives,such as virtual reality,mobile games,and mission-critical applications.The data flow of various business types is increasing exponentially,which brings huge challenges to network data transmission management and network resource management.Based on the above considerations,this thesis aims to study resource management strategies for multi-type services in SDN,including considering how to provide controller resources for different services,how to combine SDN technology and network function virtualization(NFV)technology to perceive service and establish service functions for Service Function Chain(SFC),and how to provide users with virtual resources under the premise of limited mobile edge resources and user mobility.The main contributions of this paper are as follows:1.The adaptive mapping management mechanism between the controller and the switch in the software-defined data center network with flows as the object is studied.First,a two-layer classification control mode network architecture is designed,and a new adaptive controller configuration mechanism that is consistent with the characteristics of the designed network model is proposed.Then,to satisfy the network capacity constraint and delay requirement,the mapping relationship between the controller and the switch that takes the flow as the object is established,and it is formulated as an integer programming model.To solve this problem,a low-overhead random rounding algorithm is designed,and the approximate gap between the proposed algorithm and the optimal solution is theoretically analyzed.Finally,the simulation results show that the proposed algorithm can effectively reduce the control bandwidth cost and the delay cost of routing rules in the network.2.The service-aware SFC routing and virtual resource allocation strategy in SDN is studied.First,a virtual resource configuration network model based on SDN is designed,combined with the change effect of service traffic and the characteristics of the dependency relationship between service functions,and the service-aware SFC routing problem is proposed.Then,the minimization maximum network load link problem is constructed as an integer programming model,and it is reduced to the Hamilton ring problem,proved to be NP-hard.Subsequently,three low-time complexity heuristic algorithms are proposed to solve the totally-ordered virtual SFC routing and partial-sequence SFC routing.Finally,a large number of experimental results show that the proposed algorithm is effective in reducing the packet loss rate and reducing the end-to-end transmission delay.3.The problem of VNF placement for user mobility in SDN is studied.First,in view of the problem of end-to-end delay instability of SFC caused by high-speed mobility of users,an SDN-based mobile edge network VNF placement management architecture is designed.Then,by considering user delay and edge cloud server resource capacity,a nonlinear integer programming problem for the adaptive VNF placement for mobile users is constructed by using the characteristics of user movement evaluated by SDN controller as a decision parameter.To solve this problem,an integer programming problem is obtained by using the dependency between the VNFs in the SFC and proved to be NP-hard.Subsequently,a random rounding algorithm is designed and the approximate gap between the proposed algorithm and the optimal solution is theoretically analyzed.Finally,the simulation results show that the proposed algorithm has better performance in user access rate,bandwidth consumption and CPU memory resource consumption than existing algorithms.