摘要： 针对现有的数据流测量算法所采用的对称主备切换机制存在存储空间浪费的问题，提出一种非对称主备切换机制，用于提高数据流测量算法的空间效率。非对称主备切换机制利用网络数据流达到过程的平稳特性，以及DRAM的高速块写入特性，只需采用一个小容量的备存储器，即可实现主备切换的功能。实验结果表明，与对称主备切换机制相比，非对称主备切换机制在几乎不影响测量错误概率的前提下，能够节省近一半的存储器空间。

关键词: 网络流量分析, 网络流量测量, 布鲁姆过滤器, 主备切换, 错误概率, 空间效率

Abstract: The Symmetrical Primary Standby Switching(SPSS) mechanism employed by existing flow traffic measurement algorithms can not make full use of memory space. An Asymmetrical Primary Standby Switching(APSS) mechanism is presented for improving the space efficiency of data flow measurement algorithms. The APSS mechanism is based on the observations that flow arrival process is stable, and DRAM can support bulk write which is much faster than random access. A small standby memory is enough to realize the primary-standby mechanism. Experimental results show that compared with SPSS, APSS can reduce memory consumption by almost a half while having trivial impact on the measurement error probability.

Key words: network traffic analysis, network traffic measurement, Bloom filter, primary-standby switching, error probability, space efficiency

中图分类号: 

• TP393