摘要： 为应对实际环境中网络流量的动态变化，同时降低运营商的运营成本，提出基于在线服务功能扩展的服务功能链部署方法。将空置状态虚拟功能实例的生命周期管理问题建模为雪橇租赁模型，采用雪橇租赁问题的最优解配置空置服务功能实例的生命周期，并设计自适应扩展开销最小化的在线服务功能实例扩展算法。以服务功能实例自适应扩展的结果作为约束条件，将每一时刻的服务功能链部署问题建模为整数线性规划问题，利用遗传算法求解带宽开销最小化的服务功能路径，实现动态场景下的服务功能链部署。仿真结果表明，该方法能根据网络流量变化动态调整虚拟资源，降低服务功能实例自适应扩展成本，节省带宽资源。

关键词: 服务功能链, 软件定义网络, 网络功能虚拟化, 长短时记忆, 遗传算法

Abstract: To cope with the dynamic changes of network traffic in the actual environment and reduce the running cost of operators,this paper proposes a Service Function Chain(SFC) deployment method based on online service function extension.The method firstly simplifies the life cycle management of virtual function instances in idle state to a ski-rental problem,and adopts the optimal solution of the ski-rental problem to configure the life cycle of service function instances in idle state.So an online service function instance extension algorithm that adaptively minimizes extension costs is designed.Then,with adaptive extension result of service function instances as constraints,the SFC deployment at each moment is modeled as a integer linear programming problem.The Genetic Algorithm(GA) is used to find the service function path with minimized bandwidth costs,so as to implement SFC deployment in dynamic scenarios.Simulation results show that the proposed method can dynamically adjust virtual resources based on network traffic changes,reducing bandwidth occupancy and adaptive extension costs of service function instances.

Key words: Service Function Chain(SFC), Software Defined Network(SDN), Network Function Virtualization(NFV), Long Short-Term Memory(LSTM), Genetic Algorithm(GA)

中图分类号: 

• TN915.81