Research On Deployment Algorithm Of Service Function Chain In NFV Network

One of the main directions of telecommunication network development is the innovation of network architecture.Network Function Virtualization(NFV)technology is introduced to solve the problems of low resource usage efficiency,high network operation cost,and rigid service level left over from traditional networks.NFVS,which decouple software and hardware in a network,play a crucial role in promoting architectural innovation.NFV manages the traditional intermediate box as a specific Virtual Network Function(VNF)and deploys it to the underlying physical network.By linking VNFs in an orderly way,the NFV architecture enables the deployment of end-to-end network services in a Service Function Chain(SFC)mode,which enhances the flexibility and scalability of network service deployment.While this new approach to service delivery can bring many advantages and benefits,how to deploy service functionality chains cost-effectively remains a key challenge for NFV.This thesis studies the problem,the main innovations are as follows:(1)In view of the problem that most of the existing virtual network functional tschemes only consider a single resource and ignore the multi-dimensional resource requirements of VNF,which leads to the imbalance of resource allocation,this thesis proposes an efficient multi-dimensional resource balancing VNF deployment method MRBD.In this method,a variety of resource requirements(CPU,memory,storage capacity,network bandwidth,etc.)of VNF are comprehensively considered.The server resource utilization and its threshold are expressed in the form of multi-dimensional vector,and the Euclidian distance and cosine similarity of the two are respectively used to represent the server resource usage degree and resource balance degree.Find the most appropriate deployment server for each VNF by periodically updating the resource information of the server.The method uses two optimization objectives of weight parameter combined with Euclidean distance and cosine similarity to ensure balanced resource allocation and maximize multidimensional resource utilization.This method can achieve near-optimal resource utilization in each dimension while balancing the allocation of server resources,according to the simulation results.This effectively reduces the number of servers in use compared to the existing algorithms.(2)To address the issue of excessive link resource consumption in the VNF deployment stage,which is caused by the independent deployment of VNF and link in most existing service function chain deployment schemes,this thesis proposes a two-stage service function chain deployment method MRCD-adv based on multi-dimensional resource correlation.In this method,the SFC deployment problem is regarded as a variant of the end-to-end optimal path search problem,the physical network topology is transformed into a weighted undirected graph,and the SFC request is transformed into a weighted directed graph.The joint deployment of nodes and links is used to establish a resource correlation measurement model,determine the matching degree between VNF and candidate nodes,and select the most matched candidate nodes to deploy VNF.Then,on the premise of satisfying bandwidth resource constraints,the virtual links connected to VNF are mapped to the shortest paths between corresponding nodes using k shortest path algorithm.According to the simulation results,the proposed method outperforms the existing algorithms in request acceptance rate,deployment cost and resource utilization.