Separation of timescales in a two-layered network

M. Vlasiou, J. Zhang, B. Zwart, R.D. Mei, van der

Research output: Book/ReportReportAcademic

32 Downloads (Pure)

Abstract

We investigate a network consisting of two layers occurring in, for example, application servers, and model the first layer as a many-server Jackson network. Active servers acts as customers at the second layer, where they are served by a common CPU. This system provides a benchmark example of a layered system. Our main result shows a separation of time scales in heavy traffic: the main source of randomness occurs at the (aggregate) CPU level; the interactions between different types of nodes at the other level is shown to converge to a fixed point at a faster time scale; this also yields a state-space collapse property. Apart from these fundamental insights, we also obtain an explicit approximation for the joint law of the system which is provably accurate for heavily loaded systems, and performs numerically well for moderately loaded systems. The obtained results for the model under consideration can be applied to thread-pool dimensioning in application servers.
Original languageEnglish
Place of PublicationEindhoven
PublisherEurandom
Number of pages9
Publication statusPublished - 2011

Publication series

NameReport Eurandom
Volume2011032
ISSN (Print)1389-2355

Fingerprint

Servers
Program processors

Cite this

Vlasiou, M., Zhang, J., Zwart, B., & Mei, van der, R. D. (2011). Separation of timescales in a two-layered network. (Report Eurandom; Vol. 2011032). Eindhoven: Eurandom.
Vlasiou, M. ; Zhang, J. ; Zwart, B. ; Mei, van der, R.D. / Separation of timescales in a two-layered network. Eindhoven : Eurandom, 2011. 9 p. (Report Eurandom).
@book{3104a4aa8af44fc98d2e8c35d56b761f,
title = "Separation of timescales in a two-layered network",
abstract = "We investigate a network consisting of two layers occurring in, for example, application servers, and model the first layer as a many-server Jackson network. Active servers acts as customers at the second layer, where they are served by a common CPU. This system provides a benchmark example of a layered system. Our main result shows a separation of time scales in heavy traffic: the main source of randomness occurs at the (aggregate) CPU level; the interactions between different types of nodes at the other level is shown to converge to a fixed point at a faster time scale; this also yields a state-space collapse property. Apart from these fundamental insights, we also obtain an explicit approximation for the joint law of the system which is provably accurate for heavily loaded systems, and performs numerically well for moderately loaded systems. The obtained results for the model under consideration can be applied to thread-pool dimensioning in application servers.",
author = "M. Vlasiou and J. Zhang and B. Zwart and {Mei, van der}, R.D.",
year = "2011",
language = "English",
series = "Report Eurandom",
publisher = "Eurandom",

}

Vlasiou, M, Zhang, J, Zwart, B & Mei, van der, RD 2011, Separation of timescales in a two-layered network. Report Eurandom, vol. 2011032, Eurandom, Eindhoven.

Separation of timescales in a two-layered network. / Vlasiou, M.; Zhang, J.; Zwart, B.; Mei, van der, R.D.

Eindhoven : Eurandom, 2011. 9 p. (Report Eurandom; Vol. 2011032).

Research output: Book/ReportReportAcademic

TY - BOOK

T1 - Separation of timescales in a two-layered network

AU - Vlasiou, M.

AU - Zhang, J.

AU - Zwart, B.

AU - Mei, van der, R.D.

PY - 2011

Y1 - 2011

N2 - We investigate a network consisting of two layers occurring in, for example, application servers, and model the first layer as a many-server Jackson network. Active servers acts as customers at the second layer, where they are served by a common CPU. This system provides a benchmark example of a layered system. Our main result shows a separation of time scales in heavy traffic: the main source of randomness occurs at the (aggregate) CPU level; the interactions between different types of nodes at the other level is shown to converge to a fixed point at a faster time scale; this also yields a state-space collapse property. Apart from these fundamental insights, we also obtain an explicit approximation for the joint law of the system which is provably accurate for heavily loaded systems, and performs numerically well for moderately loaded systems. The obtained results for the model under consideration can be applied to thread-pool dimensioning in application servers.

AB - We investigate a network consisting of two layers occurring in, for example, application servers, and model the first layer as a many-server Jackson network. Active servers acts as customers at the second layer, where they are served by a common CPU. This system provides a benchmark example of a layered system. Our main result shows a separation of time scales in heavy traffic: the main source of randomness occurs at the (aggregate) CPU level; the interactions between different types of nodes at the other level is shown to converge to a fixed point at a faster time scale; this also yields a state-space collapse property. Apart from these fundamental insights, we also obtain an explicit approximation for the joint law of the system which is provably accurate for heavily loaded systems, and performs numerically well for moderately loaded systems. The obtained results for the model under consideration can be applied to thread-pool dimensioning in application servers.

M3 - Report

T3 - Report Eurandom

BT - Separation of timescales in a two-layered network

PB - Eurandom

CY - Eindhoven

ER -

Vlasiou M, Zhang J, Zwart B, Mei, van der RD. Separation of timescales in a two-layered network. Eindhoven: Eurandom, 2011. 9 p. (Report Eurandom).