Abstract: Cisco’s NetFlow is widely deployed for traffic measurement，since it adopts static sampling rate，while there’s a burst of traffic，it consumes excessive router resources，and even affects the normal forward function.This paper presents a novel flow sampling method based on DCF with controllable resource.It samples a fixed number of packets in the measurement bin，and employs DCF hash algorithm to maintain flow records，which effectively controls resources consumption.Results show that the proposed method can achieve sampling rate’s self-adaptability，simplicity and controllability of resource consumption without sacrificing accuracy.

摘要： Cisco的NetFlow是流测量中广泛应用的方案，但因其采用静态抽样率，当网络流量突然上升时，消耗过多的路由器资源，甚至影响路由器的正常转发功能。提出了基于DCF的资源可控流抽样，对测量间隔内到达的报文采取固定数量的抽样，并采用DCF哈希算法维护流记录，有效控制了资源的消耗。理论和实验分析结果表明，该方法具有抽样率自适应性、简单性、资源可控性，同时不失准确性。