摘要

Discovery of service nodes in flows is a challenging task, especially in large ISPs or campus networks where the amount of traffic across network is massive. We propose an effective data structure called Round-robin Buddy Bloom Filters (RBBF) to detect duplicate elements in flows. A two-stage approximate algorithm based on RBBF which can be used for detecting service nodes from NetFlow data is also given and the performance of the algorithm is analyzed. In our case, the proposed algorithm uses about 1% memory of hash table with false positive error rate less than 5%. A prototype system, which is compatible with both IPv4 and IPv6, using the proposed data structure and algorithm is introduced. Some real world case studies based on the prototype system are discussed.