Elastic_HH: Tailored Elastic for Finding Heavy Hitters

09/05/2019

∙

by   Xilai Liu, et al.

∙

0

∙

share

Finding heavy hitters has been of vital importance in network measurement. Among all the recent works in finding heavy hitters, the Elastic sketch achieves the highest accuracy and fastest speed. However, we find that there is still room for improvement of the Elastic sketch in finding heavy hitters. In this paper, we propose a tailored Elastic to enhance the sketch only for finding heavy hitters at the cost of losing the generality of Elastic.To tailor Elastic, we abandon the light part, and improve the eviction strategy. Our experimental results show that compared with the standard Elastic, our tailored Elastic reduces the error rate to 5.7 8.1 times and increases the speed to 2.5 times. All the related source codes and datasets are available at Github.