ePrints

Multi-class resource sharing with preemptive priorities

Lookup NU author(s): Emeritus Professor Isi Mitrani

Downloads

• Accepted version [.pdf]

Licence

This is the authors' accepted manuscript of an article that has been published in its final definitive form by Cambridge University Press, 2018.

For re-use rights please refer to the publisher's terms and conditions.

Abstract

Copyright © Cambridge University Press 2017 Different virtual machines can share servers, subject to resource constraints. Incoming jobs whose resource requirements cannot be satisfied are queued and receive service according to a preemptive-resume scheduling policy. The problem is to evaluate a cost function, including holding and server costs, with a view to searching for the optimal number of servers. A model with two job types is analyzed exactly and the results are used to develop accurate approximations, which are then extended to more than two classes. Numerical examples and comparisons with simulations are presented.

Publication metadata

Author(s): Mitrani I

Publication type: Article

Publication status: Published

Journal: Probability in the Engineering and Informational Sciences

Year: 2018

Volume: 32

Issue: 3

Pages: 323-339

Print publication date: 01/07/2018

Online publication date: 11/07/2017

Acceptance date: 02/04/2016

Date deposited: 26/09/2017

ISSN (print): 0269-9648

ISSN (electronic): 1469-8951

Publisher: Cambridge University Press

URL: https://doi.org/10.1017/S0269964817000286

DOI: 10.1017/S0269964817000286

Altmetrics

Share