BF业务流查找引擎的性能研究
Research Performance of Flow Search Engine
Bloom Filter是一种数据结构,当应用于业务流查找时,更适应集中式业务流,同时更易于用硬件实现。Bloom Filter存在正向误检,对最终的查找性能有一定影响。通过软件实现Bloom Filter,同时引入缓存机制。考察不同参数条件下,Bloom Filter的误检概率和BF业务流查找引擎的性能。结果表明:存在一个最佳Hash函数的取值范围7~15,使Bloom Filter的误检概率最小;随着值阵列容量的增加,Bloom Filter的误检概率不断减小,在容量为10M的情况下,误检概率已经很低;缓存机制的引入能有效提高查找引擎的性能。
Bloom filter is a kind of data structure. When bloom filter is applied in the area of flow search, it has some advantages, such as the convenience of realization in hardware and the good performance. But the performance is affected by the characteristic called false positive. A software use bloom filter and cache is implemented. After studying the probability of false positive and the performance, the experience shows that there is a best number of hash functions which make the probability of false positive lowest, and with the increment of the bit vector, the probability of false positive decreases. If introducing cache, the performance of Bloom filter becomes better.
梁佳
计算技术、计算机技术
业务流查找Bloom Filter
Flow SearchBloom Filter
梁佳.BF业务流查找引擎的性能研究[EB/OL].(2007-09-07)[2025-06-07].http://www.paper.edu.cn/releasepaper/content/200709-128.点此复制
评论