Reconfiguration And Optimization Of NFVO Architecture Based On Microservices

With the development of network technology,the scale of the network is getting larger and larger,and the demand for flexibility is increasing.NFV(Network Function Virtualization)is a new direction for the development of network architecture.NFVO(Network Function Virtualization Orchestrator)is an important part of it,which is responsible for the basic work of orchestration,operation and maintenance.This paper investigates the open source NFVO project OPEN-O and finds problems in its existing architecture.For the architecture short board,this paper designs and implements a reconstruction scheme based on the Spring Cloud framework.In order to make full use of the advantages of the micro-service architecture,the micro-services are usually deployed on the server in the form of containers,and the micro-service container is managed by the container orchestration system,and the elastic expansion is realized by adjusting the number of micro-service copies.This paper analyzes the elastic scaling principle of Kubernetes,the most popular container orchestration system,and is based on thr-eshold-based response scaling.In order to improve system performance and enhance service stability,this paper enhances Kubernetes and designs and implements elastic scaling model based on load forecasting.At the same time,according to the shortcomings of the existing load forecasting algorithm,this paper proposes a load prediction algorithm model RNNDA based on RNN(Recurrent Neural Network)and multi-level Attention Mechanism.The algorithm comprehensively considers the influence of CPU,memory,disk and network transceiver on various loads.And added the Self-Attention and time-domain Attention Mechanism to eliminate the noise and learn the importance of the data at different times for the prediction results.Tests have verified that RNNDA has better performance than existing load prediction.This paper implements microservice reconfiguration of the network function virtualization orchestrator,design RNNDA load prediction algorithm and designs and implements the elastic scaling for Kubernetes,based on RNNDA algorithm for load prediction and increases the quality and stability of the system.