基于时变图聚合的虚拟化服务链负载均衡方法

A Load Balancing Method of Virtualization Service Function Chain Based on Time-varying Graphs Integration

针对网络虚拟化背景下云数据中心服务功能链部署所引起局部资源负载不均衡问题,提出了基于时变图聚合的服务链负载均衡方法.通过将网络划分为一系列静态时变图,来描述时间窗内数据中心网络部署服务链的动态性,将静态时变图在时间序列上排列,并在相邻时变图上的同一节点之间建立链路,使之聚合为二维,从而将虚拟资源的分配问题转化为多商品流问题,并通过启发式算法高效地解决这一问题.实验结果表明,该方法与传统服务链部署方法相比,能有效降低最大链路利用率,并降低用户数据包的平均时延.

Concerning load unbalanced caused by service function chain deployment in cloud data center by network function virtualization,a load balancing method of service chain based on time-varying graphs integration were proposed. In order to represent the dynamic network of service function chain deployment in data center within time span by dividing dynamic network into a series of static time-varying graphs, integrating static time-varying graphs into multi-dimensional along time series,establishing link between the same node in adjoined time-varying graphs,finally transform the virtual resource allocation problem into multi-commodity network flow model. This paper also developed an efficient heuristic algorithm to effectively solve the problem. Compared to the previous generally used service function chain deploying method,the simulation results show that the proposed method can substantially not only lower the performance of max physical link utilization,but also lower the average latency of tenant packets.