Research On The Construction And Optimization Of Virtual Network Function Service Chain Oriented To Microservice Architecture

With the rapid development of networks and the continuous expansion of business scale,the traditional network model has been far from meeting the increasingly diverse needs of users.The emergence of Network Function Virtualization(NFV)has promoted changes in network technology.However,most NFV technologies usually replace monolithic hardware with monolithic VNFs.Many VNFs have common functions(such as header parsing,packet classification,etc.),which will cause unnecessary resource consumption and operation overhead when dealing with these repeated common functions.Microservices have the feature of decoupling.By decomposing a monolithic VNF into independently deployable microservice components,the VNF can be flexibly decoupled and deployed.Aiming at the problem of duplication of network functions and excessive load of underlying network nodes,this thesis proposes a dynamic decouple and merge algorithm for VNF based on microservices.The algorithm can dynamically decouple and merge according to the current network conditions and service requirements to alleviate the excessive load of the underlying physical nodes.The experimental results show that the algorithm can reduce the delay by 34% and improve the maximum processing bandwidth of the service chain by 43% according to the load change.The number of hops that data flows through nodes is one of the important factors of network delay.Therefore,on the basis of decoupling and merging,this thesis proposes a shortest path-oriented functional service chain deployment algorithm for the further optimization of delay and bandwidth.By searching for the shortest path,the data flow will pass through fewer physical nodes,thereby further optimizing the delay.At the same time,combining the packet loss rate and delay,the flow table can be reordered to reduce the burden of the service chain and alleviate the congestion of the link.According to the experimental results,this algorithm can greatly reduce the delay by 31%-21%.In terms of bandwidth utilization,in an environment with a packet loss rate of 7%-15%,the bandwidth utilization rate can be increased by approximately 3%-6%.