| With the rapid development of the Internet,the traditional way of providing network services has been difficult to meet the growing network needs.Network Functions Virtualization(NFV)is an emerging network architecture used to solve the difficulties encountered in the traditional way such as high cost and difficult management,so that network functions can be implemented in software on general hardware.One of the main challenges in the research of Network Function Virtualization is the allocation of network resources and how to reasonably arrange and schedule virtual network functions is one of the keys.The problem of orchestration and scheduling of virtual network functions is very challenging.In NFV environment,network service is a set of linked virtual network functions.We need to allocate virtual network functions and virtual links in a reasonable way and execute them according to requirements.How to get a better way to execute virtual network functions is the focus of the research of orchestration and scheduling problems.We need to ensure that the service performance such as reliability and security will not be reduced while minimizing the end-to-end delay and it follows all priorities and dependencies between the virtual network functions that make up the network service.This is also the key to ensuring the quality of network services and the rational allocation of resources.This paper aims to solve the problem of orchestration and scheduling of virtual network functions in different network service scenarios in the NFV environment.This paper proposes reasonable solutions to allocate network resources to reduce the end-to-end delay of network services and ensure the quality of network services.In summary,the main research contents of this article include:(1)An offline virtual network function scheduling method based on improved genetic algorithm is proposed.This topic uses the improved scheme of genetic algorithm to schedule virtual network functions of network services in offline network scenarios and gives the specific implementation of each virtual network function in the substrate node.This solution aims at optimizing the overall end-to-end delay of network services.A problem model that considers processing delay and transmission delay is established.Then three improvements are given for the genetic algorithm.Finally,the improved algorithm is applied in the problem scenario of offline virtual network functions.The provision of offline network service and the optimization of resource are realized.(2)An online virtual network function scheduling method based on particle swarm optimization is proposed.In order to deal with network services that arrive in real-time,this solution will provide mapping and scheduling schemes for network services under the constraints of substrate resources and network services.It determines mapping locations for virtual network functions and gives reasonable and correct scheduling timing.This paper first determines the goal of optimizing the weighted average end-to-end delay of real-time network services in an online scenario and proposes a problem model that considers the priority of network service and satisfies the constraints of mapping and scheduling.Then we consider the characteristics of real-time network services and give the idea of adjusting the scheduling scheme in real-time according to the arrival of network service requests.Finally,the particle swarm algorithm is used to calculate the best online scheduling scheme to ensure the real-time service quality of the network. |
