基于两层自适应超时策略的资源可控流抽样  

Flow Sampling of Controllable Resource Based on Two-Layer Self-Adaptive Timeout Strategy

作　　者：孟金凤[1] 高仲合[1] 禹继国[1] 

机构地区：[1]曲阜师范大学计算机科学学院,山东日照276826

出　　处：《济南大学学报（自然科学版）》2014年第5期371-375,共5页Journal of University of Jinan(Science and Technology)

基　　金：国家自然科学基金(61373027);山东省自然科学基金(ZR2012FM023)

摘　　要：NetFlow是流量测量中广泛使用的技术,但其采用固定的抽样率,在流量较大时消耗过多的系统资源,而在流量较小时造成资源利用不足,且在实际应用中缺乏一定的资源保护机制。为解决其缺陷,提出一种基于两层自适应超时策略的资源可控流抽样测量方法。该方法首先根据时间进行分层,对定长单位时间内的报文进行固定数量的随机抽样,并对其进行流统计,最后用两层自适应超时策略控制流的输出。理论分析和实验表明,该抽样方法不仅具有准确性、简单性和资源可控性,而且从很大程度上提高了"流cache"的利用率。Netflow is a widely used technology in flow measurement, however, due to adopting fixed sampling rate, it consumes too many system resources when the flow is large and causes insufficient resource utilization when the flow is small, and it has no a certain resource protection mechanism in practiee. In order to solve these defects, we put forward a kind of sampling measurement method based on the two-layer self-adaptive timeout strategy. At first we stratify the flow according to time, then randomly sample a fixed number of messages in a fixed time unit and count the flow, and finally control the flow output with the two-layer adaptive timeout strategy. It is shown that this sampling method is accurate, simple and controllable for resources, and greatly improves the utilization rate of the ＂flow cache＂.

关 键 词：流抽样 自适应超时 时间分层 资源可控 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]