Design And Implementation Of Virtual Function Elastic Adaptation Mechanism Under Smart Integration Identifier Networking

With the diversification of network application scenarios and user services,the current Internet has been difficult to meet the increasing demands for customized services.However,the existing network function deployment and migration mechanism are relatively static and rigid,so it is difficult to solve the tight coupling problem between network and service.In this paper,based on the Smart Integration Identifier Networking architecture,SDN and NFV technologies are integrated to study the elastic adaptation of virtual network functions.This paper aims to realize the dynamic migration of virtual network functions on demand and improve the scalability and flexibility of the network.The specific work of this paper is as follows:Firstly,this paper analyzes the design requirements of the elastic adaptation mechanism of virtual functions,uses the node load and delay of the Service Function Chaining as the evaluation indicators,proposes an elastic adaptation mechanism of virtual functions under the Smart Integration Identifier Networking,designs a virtual function elastic adaptation system based on the node function migration.The system consists of six functional modules: the service management module provides API interface,receives and analyzes the service demand;the network and resource awareness module is based on ONOS controller and Kubernetes cloud platform to dynamically perceive the resource information;the data storage module records and maintains the information of the whole network;the demand and resource monitoring module monitors and analyzes the state of network;the elastic adaptation module judges the time when the network node triggers the virtual network function migration and makes the corresponding migration decision;the deployment issuing module executes the specific migration policies through the orchestrator and the controller.Aiming at the decision requirements of virtual network function migration,this paper proposes a decision optimization algorithm of the virtual network function migration based on deep reinforcement learning.Based on the constraints of network and node resources,this algorithm establishes a migration decision model that includes state space,behavior space,and reward mechanism,and realizes the elastic adaptation of virtual network functions.The migration decision includes two stages: offline training and online decision-making.In the offline training stage,deep reinforcement learning is used to train the strategy network model;the online migration stage is based on the trained strategic network model,which takes the current network state as the input of the model,dynamically generates the migration strategy,and performs the migration of virtual network functions online.The algorithm optimizes the load of the node in the network.,and solves the problem of service interruption caused by overloading of a single node and failure of a link and virtual function instances.Finally,this paper tests the function and performance of each module of the virtual function elastic adaptation mechanism.Functional test results show that the mechanism can coordinate each module to realize the discovery and update of service demand,dynamically sensing and monitoring the network resources,timely respond to the problems such as the overloading of node and failure of virtual network function instances,complete virtual function migration,and avoid the interruption of service.In terms of performance test,this paper applies the migration algorithm in the adaptation mechanism to the prototype system of Smart Integration Identifier Networking and compares node load and the end-to-end delay with the random migration algorithm and the shortest path migration algorithm,the experimental results verify the effectiveness of the proposed algorithm.