面向NFV中间盒依赖关系的有效路由策略

Effective routing strategy based on NFV middleboxes dependency

在软件定义网络中,通过网络功能虚拟化(network function virtualization,NFV)可以有效地优化中间盒的部署以及数据包的路由,但是受中间盒之间存在的依赖关系约束,不适当的部署策略会为运营商带来额外的路由成本,降低资源利用率。为此,文章建立混合整数线性规划模型,提出一种服务链感知精准算法用于计算受中间盒依赖关系以及链路带宽约束的最小路由成本。该算法首先基于中间盒的依赖关系构造有向层级图,然后从该层级图中的源点到终点之间的最短路径中筛选出满足带宽要求的链路作为数据包的路由。仿真实验结果表明,该算法可以快速获取最优解,具有较好的应用前景。

Network function virtualization(NFV)can effectively optimize the deployment of middleboxes and the routing of network flows in Software Defined Network.However,due to the constraints existing between middleboxes,improper deployment strategies will bring operators additional routing costs and reduce management efficiency.To this end,this paper proposes a service chain aware exact algorithm(SCAEA)based on the mixed integer linear programming(MILP)to minimize routing costs under the constraints of middlebox dependencies and link bandwidth.The algorithm first constructs a directed hierarchical graph based on the dependency of the middleboxes,and then uses the shortest path from source to destination as the routing path in the hierarchical graph.Simulation results show that the algorithm can quickly obtain accurate solutions and is promising.