基于批量计算的网络状态细粒度感知方法

Fine Particle Perception of Network State Based on Batch Computing

随着网络用户数量的不断增多以及网络规模的不断扩大,各种新型网络应用不断涌现,传统的网络形态无法及时满足这些新业务和新需求,给当前的网络体系结构和运行方式带来了挑战。传统网络的不足之处,比如升级成本高、运维管理方式复杂以及业务创新缓慢等问题暴露得越来越明显。针对在高速网络测量中无法兼顾测量精度、覆盖面和计算成本的问题,提出了一种可部署在硬件交换机上的批量计算测量报告生成方法,将原来一次性生成报告分成多次计算生成报告。该方法能够在Tbit/s级流量下实现精确的网络测量,并且只需要消耗很低的硬件资源,既保障了测量指标的丰富程度,又降低了计算成本。

With the increasing number of network users and the continuous expansion of network scale,a variety of new network applications continue to emerge,and the traditional network form cannot timely meet these new business and new needs,which brings challenges to the current network architecture and operation mode.The shortcomings of traditional network,such as high upgrade cost,complex operation and maintenance management,slow business innovation and other problems exposed more and more clearly.To solve the problem that measurement precision,coverage and calculation cost cannot be taken into account in high-speed network measurement,a method of generating batch calculation and measurement report which can be deployed on hardware switch is proposed.This method can achieve accurate network measurement under Tbps level traffic with very low hardware resources,which not only ensures the richness of measurement indicators,but also reduces the calculation cost.