Adaptive fair sampling method based on reduction of number of flows

Abstract

The invention relates to an adaptive fair sampling method based on reduction of number of flows, belonging to the field of network traffic measurement. The adaptive fair sampling method comprises: according to whether the arrival packet is an existing flow table entries, different network flow fairness sampling strategies are obtained; according to the reduction of number of flows, the size flow is distinguished and counted, and the selective extraction ratio is obtained, and the flow table entries in the memory cache are established; according to the speed of new flow table entries arriving at the measure point, all sample flow sets with flow integer compression are obtained; a new sampling probability function cluster is proposed based on the size flow distribution of all sample flow sets; according to the probability function cluster, the sample stream set is sampled fairly, and the fair sampling results of the size flow in the sample are obtained. The adaptive fair sampling method based on reduction of number of flows can realize the accuracy of the sampling algorithm in the network flow measurement. And the adaptive fair sampling method based on reduction of number of flows can not only solve the scalability problem of the sampling algorithm on the high-speed chain path, but also effectively improve the fairness of the algorithm.

Description

technical field [0001] The invention belongs to the field of network flow measurement, in particular to an adaptive fair sampling guarantee method based on flow number reduction. Background technique [0002] Network traffic measurement quantifies the various indicators of the flow, intuitively describes the composition of the current network traffic, and reflects the current operating status of the network. It plays an extremely important role in applications such as traffic billing, traffic identification, fault detection, and network security. . Since the growth rate of data on the network far exceeds the growth rate of memory performance, real-time statistics of each flow has become a huge problem in real-time traffic measurement of high-speed backbone networks. Data compression through sampling is an important means of real-time measurement of high-speed networks. However, existing sampling algorithms improve the accuracy of small streams at the cost of sacrificing the...

AI Technical Summary

Problems solved by technology

[0004] The present invention aims at improving the accuracy of small streams at the cost of sacrificing the accuracy of large streams in the existing sampling algorithm, which leads to weak fairness of the algorithm and cannot well solve problems such as the scalability of the algorithm, and proposes a An Adaptive Fair Sampling Guarantee Method Based on Flow Number Reduction

Images