Algorithms and estimators for summarization of unaggregated data streams

Edith Cohen; Nick Duffield; Haim Kaplan; Carstent Lund; Mikkel Thorup

doi:10.1016/j.jcss.2014.04.009

Algorithms and estimators for summarization of unaggregated data streams

Edith Cohen, Nick Duffield, Haim Kaplan, Carstent Lund, Mikkel Thorup

Department of Computer Science

10 Citations (Scopus)

Abstract

Statistical summaries of IP traffic are at the heart of network operation and are used to recover aggregate information on subpopulations of flows. It is therefore of great importance to collect the most accurate and informative summaries given the router's resource constraints. A summarization algorithm, such as Cisco's sampled NetFlow, is applied to IP packet streams that consist of multiple interleaving IP flows. We develop sampling algorithms and unbiased estimators which address sources of inefficiency in current methods. First, we design tunable algorithms whereas currently a single parameter (the sampling rate) controls utilization of both memory and processing/access speed (which means that it has to be set according to the bottleneck resource). Second, we make a better use of the memory hierarchy, which involves exporting partial summaries to slower storage during the measurement period.

Original language	Danish
Journal	Journal of Computer and System Sciences
Volume	80
Issue number	7
Pages (from-to)	1214-1244
Number of pages	31
ISSN	0022-0000
DOIs	https://doi.org/10.1016/j.jcss.2014.04.009
Publication status	Published - Nov 2014

Access to Document

10.1016/j.jcss.2014.04.009

Cite this

@article{e63bee21fa3b4a248cdaa9c82abeff76,

title = "Algorithms and estimators for summarization of unaggregated data streams",

abstract = "Abstract Statistical summaries of IP traffic are at the heart of network operation and are used to recover aggregate information on subpopulations of flows. It is therefore of great importance to collect the most accurate and informative summaries given the router's resource constraints. A summarization algorithm, such as Cisco's sampled NetFlow, is applied to IP packet streams that consist of multiple interleaving IP flows. We develop sampling algorithms and unbiased estimators which address sources of inefficiency in current methods. First, we design tunable algorithms whereas currently a single parameter (the sampling rate) controls utilization of both memory and processing/access speed (which means that it has to be set according to the bottleneck resource). Second, we make a better use of the memory hierarchy, which involves exporting partial summaries to slower storage during the measurement period.",

keywords = "NetFlow, Data streams, Random sampling, IP flows, Subpopulation queries, Flow size distribution",

author = "Edith Cohen and Nick Duffield and Haim Kaplan and Carstent Lund and Mikkel Thorup",

year = "2014",

month = nov,

doi = "10.1016/j.jcss.2014.04.009",

language = "Dansk",

volume = "80",

pages = "1214--1244",

journal = "Journal of Computer and System Sciences",

issn = "0022-0000",

publisher = "Academic Press",

number = "7",

}

TY - JOUR

T1 - Algorithms and estimators for summarization of unaggregated data streams

AU - Cohen, Edith

AU - Duffield, Nick

AU - Kaplan, Haim

AU - Lund, Carstent

AU - Thorup, Mikkel

PY - 2014/11

Y1 - 2014/11

N2 - Abstract Statistical summaries of IP traffic are at the heart of network operation and are used to recover aggregate information on subpopulations of flows. It is therefore of great importance to collect the most accurate and informative summaries given the router's resource constraints. A summarization algorithm, such as Cisco's sampled NetFlow, is applied to IP packet streams that consist of multiple interleaving IP flows. We develop sampling algorithms and unbiased estimators which address sources of inefficiency in current methods. First, we design tunable algorithms whereas currently a single parameter (the sampling rate) controls utilization of both memory and processing/access speed (which means that it has to be set according to the bottleneck resource). Second, we make a better use of the memory hierarchy, which involves exporting partial summaries to slower storage during the measurement period.

AB - Abstract Statistical summaries of IP traffic are at the heart of network operation and are used to recover aggregate information on subpopulations of flows. It is therefore of great importance to collect the most accurate and informative summaries given the router's resource constraints. A summarization algorithm, such as Cisco's sampled NetFlow, is applied to IP packet streams that consist of multiple interleaving IP flows. We develop sampling algorithms and unbiased estimators which address sources of inefficiency in current methods. First, we design tunable algorithms whereas currently a single parameter (the sampling rate) controls utilization of both memory and processing/access speed (which means that it has to be set according to the bottleneck resource). Second, we make a better use of the memory hierarchy, which involves exporting partial summaries to slower storage during the measurement period.

KW - NetFlow

KW - Data streams

KW - Random sampling

KW - IP flows

KW - Subpopulation queries

KW - Flow size distribution

U2 - 10.1016/j.jcss.2014.04.009

DO - 10.1016/j.jcss.2014.04.009

M3 - Tidsskriftartikel

SN - 0022-0000

VL - 80

SP - 1214

EP - 1244

JO - Journal of Computer and System Sciences

JF - Journal of Computer and System Sciences

IS - 7

ER -