The effect of traffic self-similarity on network performance

Park
Kim
Crovella

Abstract

Recent measurements of network traffic have shown that self-similarity is an ubiquitous phenomenon present in both local area and wide area networks. In previous work, we have shown a simple, robust application layer causal mechanism of traffic self-similarity, namely, the transfer of objects or files in a generic network system where the file size distributions are heavytailed. In this paper, we study the effect of scale-invariant burstiness on network performance when the protocol stack and the interaction of traffic sources sharing bounded network resources are incorporated. First, we present a framework for understanding the two principal performance implications of self-similar traffic---concentrated periods of congestion and amplified packet loss/queueing delay. Using performance

Keywords

self similar
burstiness
heavy tail
scale-invariant
congestion

Notes

Park, et al, recreate the observed self similarity and scale invariance of real traffic using the ns network simulator with UDP and TCP traffic. They show that the simple transfer of files with a long tailed size distribution can give the observed behaviour. They study the effect of flow control on this bursty traffic.

Related Papers

Bibtex

 @inproceedings{ park_kim.selfsimilar97,
    author = "K. Park and G. Kim and M. Crovella",
    title = "The effect of traffic self-similarity on network performance",
    booktitle = "Proc. SPIE International Conference on Performance and Control of Network Systems, November",
    year = "1997"
}

Back to Intro By Author By Importance By Keyword By Title By Reference