Research On Deployment Strategy And Selection Mechanism Of Virtualized Network Function In Data Center Network

In traditional network architecture,network functions are realized in dedicated hardware.The emergence of new services makes the traditional network be facing several challenges,such as increased capital investment and slow deployment of services.Network Function Virtualization(NFV)technology enables the network functions to be realized in the form of software.By implementing Virtualized Network Functions(VNFs)in data center,the hardware cost can be greatly reduced and the deployment cycle is also shortened.However,the introduction of NFV technology leads to the problem of VNF deployment and selection in the data center networks.It is vital important to complete the VNF deployment and selection based on current network status and service requirement.This thesis focuses on the problem of VNF deployment and selection in data center networks and studies the control architecture,deployment strategy and selection mechanism,respectively.The main contributions and innovations are as follows:(1)Considering the collaboration between application resources in data center and spectrum resources in optical networks,a centralized control architecture is designed to realize the unified and programmable control in data center networks.Meanwhile,the process of VNF deployment is also analyzed in this control architecture.(2)The problem of VNF deployment in the data center network is investigated from the perspective of network survivability.Two different strategies of VNF deployment are proposed based on whether network failures are known or not.Simulation is conducted to verify the performance of the strategies.Simulation results show that the proposed algorithm can effectively reduce the impact of node failures on the VNF deployment in the data center network.When the number of failure nodes is 2,the proposed algorithm reduces the average number of invalid VNFs by 90%compared with benchmark algorithm.(3)For the problem of VNF selection,the concept of joint balancing factor is proposed in terms of load balancing of application resources and spectrum resources.A novel Joint-Optimization Selection Algorithm(JOSA)based on Local Balancing factor(LB-factor)and Global Balancing factor(GB-factor)is designed,and simuation testbed is set up in terms of two cases:different traffic loads and different ratios between application resource and spectrum resource.Simulation results show that compared with the benchmark algorithm,the blocking probability of JOSA based on GB-factor is reduced by 7%in different traffic loads,and the utilization ratio of application resource is also increased by about 9%.