基于LEAST的高速网络大流检测算法  

An Elephant Flow Identifying and Measuring Algorithm Based on LEAST in High-speed Network Environment

作　　者：徐敏[1] 夏靖波[1] 申健[1] 陈珍[1] 

机构地区：[1]空军工程大学信息与导航学院,西安710077

出　　处：《空军工程大学学报（自然科学版）》2015年第4期62-65,共4页Journal of Air Force Engineering University(Natural Science Edition)

基　　金：陕西省自然科学基金资助项目(2012JZ8005)

摘　　要：针对大流漏检率过高，占用SRAM过大问题，提出了基于最少（LEAST）改进型大流检测算法。主要思想：利用LEAST淘汰机制将小流丢弃使得大流能够被保护，采用窗口-储备策略解决检测大流的公平性问题。通过相关组织所提供的实际互联网数据进行了实验比较，结果显示：与现有算法相比，新算法具有更高的测量准确性，平均大流漏检率降低至0％～0．13％。In high-speed network environment, ifs very important to extract elephant flow timely and accu- rately for cognizing behavior and law of network. In order to reduce the elephant flow measurement miss- ing rate and overmuch occupation of SRAM, an improved algorithm based on LEAST is proposed. By u- sing LEAST elimination mechanism for discarding the mice flow, the elephant flow can be protected. And Window-Reserve strategy is adopted to ensure the fairness of identifying and measuring elephant flow. Fi- nally, through the comparison between the simulation results and the actual flow data, the result shows that the new algorithm has a higher measurement accuracy and is more practicable, and the elephant flow on the average measurement missing rate is reduced to 0%-0.13%.

关 键 词：网络测量 大流流量 LEAST淘汰机制 窗口-储备策略 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]