Fitting heavy-tailed HTTP traces with the new stratified EM-algorithm

Research output: Other contributionOther research output

Abstract

A typical step in the model-based evaluation of communication systems is to fit measured data to analytically tractable distributions. Due to the increased speed of today's networks, even basic measurements, such as logging the requests at a Web server, can quickly generate large data traces with millions of entries. Employing complex fitting algorithms on such traces can take a significant amount of time. In this paper, we focus on the Expectation Maximization-based fitting of heavy-tailed distributed data to hyper-exponential distributions. We present a data aggregation algorithm which accelerates the fitting by several orders of magnitude. The employed aggregation algorithm has been derived from a sampling stratification technique and adapts dynamically to the distribution of the data. We illustrate the performance of the algorithm by applying it to empirical and artificial data traces.
Original languageEnglish
PublisherIEEE Computer Society
Number of pages8
Place of PublicationLos Alamitos
ISBN (Print)978-1-4244-1845-9
DOIs
Publication statusPublished - Feb 2008
Externally publishedYes

Fingerprint

HTTP
Agglomeration
Communication systems
Servers
Sampling

Keywords

  • EWI-12273
  • IR-64721
  • METIS-250951

Cite this

Sadre, Ramin ; Haverkort, Boudewijn R. / Fitting heavy-tailed HTTP traces with the new stratified EM-algorithm. 2008. Los Alamitos : IEEE Computer Society. 8 p.
@misc{32778d26bfe349929d44bdbfe550e83c,
title = "Fitting heavy-tailed HTTP traces with the new stratified EM-algorithm",
abstract = "A typical step in the model-based evaluation of communication systems is to fit measured data to analytically tractable distributions. Due to the increased speed of today's networks, even basic measurements, such as logging the requests at a Web server, can quickly generate large data traces with millions of entries. Employing complex fitting algorithms on such traces can take a significant amount of time. In this paper, we focus on the Expectation Maximization-based fitting of heavy-tailed distributed data to hyper-exponential distributions. We present a data aggregation algorithm which accelerates the fitting by several orders of magnitude. The employed aggregation algorithm has been derived from a sampling stratification technique and adapts dynamically to the distribution of the data. We illustrate the performance of the algorithm by applying it to empirical and artificial data traces.",
keywords = "EWI-12273, IR-64721, METIS-250951",
author = "Ramin Sadre and Haverkort, {Boudewijn R.}",
year = "2008",
month = "2",
doi = "10.1109/ITNEWS.2008.4488162",
language = "English",
isbn = "978-1-4244-1845-9",
publisher = "IEEE Computer Society",
address = "United States",
type = "Other",

}

Fitting heavy-tailed HTTP traces with the new stratified EM-algorithm. / Sadre, Ramin; Haverkort, Boudewijn R.

8 p. Los Alamitos : IEEE Computer Society. 2008, .

Research output: Other contributionOther research output

TY - GEN

T1 - Fitting heavy-tailed HTTP traces with the new stratified EM-algorithm

AU - Sadre, Ramin

AU - Haverkort, Boudewijn R.

PY - 2008/2

Y1 - 2008/2

N2 - A typical step in the model-based evaluation of communication systems is to fit measured data to analytically tractable distributions. Due to the increased speed of today's networks, even basic measurements, such as logging the requests at a Web server, can quickly generate large data traces with millions of entries. Employing complex fitting algorithms on such traces can take a significant amount of time. In this paper, we focus on the Expectation Maximization-based fitting of heavy-tailed distributed data to hyper-exponential distributions. We present a data aggregation algorithm which accelerates the fitting by several orders of magnitude. The employed aggregation algorithm has been derived from a sampling stratification technique and adapts dynamically to the distribution of the data. We illustrate the performance of the algorithm by applying it to empirical and artificial data traces.

AB - A typical step in the model-based evaluation of communication systems is to fit measured data to analytically tractable distributions. Due to the increased speed of today's networks, even basic measurements, such as logging the requests at a Web server, can quickly generate large data traces with millions of entries. Employing complex fitting algorithms on such traces can take a significant amount of time. In this paper, we focus on the Expectation Maximization-based fitting of heavy-tailed distributed data to hyper-exponential distributions. We present a data aggregation algorithm which accelerates the fitting by several orders of magnitude. The employed aggregation algorithm has been derived from a sampling stratification technique and adapts dynamically to the distribution of the data. We illustrate the performance of the algorithm by applying it to empirical and artificial data traces.

KW - EWI-12273

KW - IR-64721

KW - METIS-250951

U2 - 10.1109/ITNEWS.2008.4488162

DO - 10.1109/ITNEWS.2008.4488162

M3 - Other contribution

SN - 978-1-4244-1845-9

PB - IEEE Computer Society

CY - Los Alamitos

ER -