Fault Detection And Management In NFV Environment

Network function virtualization is currently a hot research direction.It standardizes and improves the existing network,adopts a method of decoupling software and hardware,and uses software to replace traditional professional equipment on standard x86 hardware,thereby achieving various network functions.It improves the efficiency of network function development and makes it easier to deploy and manage network functions.Although network function virtualization is currently being used in a large number of applications,it is more unstable than traditional professional equipment,which brings considerable challenges to network stability.The network in the network function virtualization environment is more complicated than the traditional network.At the same time,faults are more likely to occur due to the impact of software.After a fault occurs,it is very important to be able to quickly locate the fault and perform corresponding maintenance on the system to provide long-term reliable and stable services.In this paper,the two aspects of fault detection and fault management in the network function virtualization environment are researched accordingly,and the fault prediction and advance processing technology are supplemented.(1)Aiming at the defects of traditional detection technology in the network function virtualization environment,this paper uses a fusion model based on machine learning,and proposes a technical process for fault detection.This method has good adaptability to the network function virtualization environment,and can improve the defects that traditional detection cannot make up,and improve the accuracy of detection.At the same time,a certain correction mechanism has been adopted to prevent misjudgments caused by network fluctuations.After that,we nalyze the cause of the fault and the location of the fault.Ensure that the source and cause of the fault can be found.(2)In terms of fault management.We first proposed a method for fault prediction processing,using gcForest as a model for fault prediction.According to the label data obtained above,the network status is detected in real time,and the upcoming network failure is predicted in advance,which is convenient for us to deal with the network problem in advance,enhance the stability of the network,and reduce the problem of service suspension due to service quality.On this basis,we have proposed a network management model that uses different management methods for network nodes of different severity to reduce the overhead caused by node migration and deployment.This paper also studies the migration and deployment of faulty nodes,proposes a multiobjective decision-making migration node selection algorithm,finds the most suitable migration target to migrate the network nodes,minimizes the migration overhead,and balances the network load.To a certain extent increases the stability of the network.This paper simulates the network architecture in the NFV environment,and experiments with different types of faults injected into the VNF.The simulation results show that the series of mechanisms proposed in this article can effectively detect and locate faults,and timely detect faults and upcoming faults.The problem is fixed.