Design And Implementation Of Intelligent SFC Orchestration System In Smart Integration Identifier Networking

With the continuous growth of user scale and the continuous extension of application scenarios,the traditional Internet has gradually exposed its shortcomings in terms of security,flexibility,resource utilization,and management/control capabilities,making it difficult to support the high-speed and high-quality development of new Internet businesses.In view of the existing problems,Smart Integration Identifier Networking(SINET)introduces the design concept of hierarchical domain and knowledge-driven,and intelligently integrates multi-space and multi-dimensional resources of the entire network,providing new ideas for the innovation of the network architecture.In order to carry out centralized management,resource optimization,and flexible deployment of today’s large-scale network devices and services,this paper conducts in-depth research on the intelligent orchestration of Service Function Chaining(SFC)based on SINET.This paper aims to achieve efficient,flexible and dynamic virtual network function deployment,endow the network with intelligent and automated service function chain management and control capabilities,and improve network resource utilization.This paper designs and implements a service function chaining intelligent orchestration system in the SINET scenario.The system separates control and forwarding,decouples software and hardware,and can receive service demand behavior descriptions and realize automatic deployment of SFC.At the same time,a reinforcement learning based SFC embedding algorithm is proposed,which comprehensively considers the delay and resource utilization to optimize resource allocation problems.The main work of this paper is as follows:Firstly,this paper summarizes the research status of NFV and SFC at home and abroad,briefly describes the architecture of SINET,and introduces concepts and technologies such as reinforcement learning and Intent-Based Network(IBN).Secondly,this paper analyzes the design requirements of the SFC intelligent orchestration system,and gives the overall design and the design of each module,including service management module,service engine module,knowledge base module,data perception and storage module,orchestration and deployment module and network components layer module,the design ideas,working process and specific implementation of each module are introduced in detail.In addition,aiming at the SFC resource allocation problem,this paper establishes the Markov Decision Process(MDP)model,designs the optimization objective,state space,action space and reward function,etc.,and proposes a lightweight SFC embedding algorithm based on Q-learning to optimize the delay and resource utilization.Finally,this paper builds an experimental topology in the test environment to test the function and performance of the system and algorithm,and verifies the effectiveness of modules such as service management,service engine,knowledge base,data perception and storage,and orchestration and deployment.At the same time,the algorithm proposed in this paper and three comparison algorithms are tested in a large-scale simulation environment.Considering different indicators such as acceptance ratio and average delay,it is verified that the algorithm has better performance.