Research On Reliability Deployment And Migration Algorithms Of Service Function Chain Based On NFV

With the rapid development of next generation network technology,massive terminals access the network,the traditional architecture can not meet the demand.Service providers introduced Network Function Virtualization(NFV)to transform the network architecture.Network function virtualization decouples network functions from hardware,deploying software on a general-purpose server to implement network functions on legacy dedicated hardware.In the NFV scenario,the service function chain serves as a basic service model to enable flexible deployment of services,provide customized network services,and support elastic scalability and load balancing of services.The main focus of this thesis is on the reliability deployment and migration issues of the service function chain.In response to the above research content,this thesis mainly carries out the following three aspects:(1)A reliability network deployment algorithm based on backup revenue index and remapping is proposed for service chain reliability.The service function chain reliability problem is transformed based on the voting model.The mathematical model is established for the reliability problem of the service function chain in the instantiation stage and the fault stage.It is proposed to use shared backup,network function re-mapping and reliability compensation mechanism to solve the reliability problems in each stage.On the premise of satisfying the reliability requirement,the backup overhead can be reduced to the greatest extent and the long-term reliability of the service function chain can be guaranteed.Simulation results show that the reliability deployment algorithm based on dual shared backup can effectively improve the reliability of service function chain.(2)In order to realize the flexible deployment of services and solve the problem of network load imbalance,a fast adaptive migration algorithm based on multi-dimensional environment perception for virtual network functions is proposed.Through resource-aware algorithms and multi-objective decision-making algorithms,load balancing of network resources is realized while ensuring business continuity.(3)Aiming at the trade-off between backup overhead and reliability fluctuation in reliability algorithm,a simulation environment is built to verify the effectiveness of the algorithm.This thesis compares the performance of different aspects of various algorithms,and gives the corresponding best solution according to their respective performance requirements in different scenarios.