Independent counter estimation buckets

摘要: Measurement capabilities are essential for a variety of network applications, such as load balancing, routing, fairness and intrusion detection. These require large counter arrays in order to monitor the traffic all flows. While commodity SRAM memories capable operating at line speed, they too small accommodate arrays. Previous works suggested estimators, which trade precision reduced space. However, accurately estimate largest counter, these methods compromise accuracy rest counters. In this work we present closed form representation optimal estimation function. We then introduce Independent Counter Estimation Buckets (ICE-Buckets), novel algorithm that improves This is achieved by separating flows buckets configuring function according each bucket's scale. prove an improved upper bound on relative error demonstrate improvement up 57 times real Internet packet traces.