Regenerative exact simulation for multiserver job model

Perfect/exact sampling has been used in situations, where the underlying distribution from which random samples are of interest is unknown or complex, for more than three decades now. However, to apply perfect sampling techniques to specific models, additional techniques are required. In this paper,...

Full description

Saved in:
Bibliographic Details
Published in:Cluster computing Vol. 28; no. 14; p. 937
Main Authors: Golovin, Alexander, Rumyantsev, Alexander, Chakravarthy, Srinivas
Format: Journal Article
Language:English
Published: New York Springer US 01.11.2025
Springer Nature B.V
Subjects:
Algorithms
Computer Communication Networks
Computer Science
Equilibrium
Expected values
Operating Systems
Processor Architectures
Queuing theory
Sampling methods
Simulation
Supercomputers
Workloads
Workload vector
65C20
Perfect sampling
Multiserver job model
Stochastic monotonicity
Kiefer–Wolfowitz recursion
Busy cycle
60K25
ISSN:1386-7857, 1573-7543
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Perfect/exact sampling has been used in situations, where the underlying distribution from which random samples are of interest is unknown or complex, for more than three decades now. However, to apply perfect sampling techniques to specific models, additional techniques are required. In this paper, the perfect sampling technique is adapted to study a workload process as well as the per-class waiting times of customers in a computing cluster model (also known as the multiserver job model) by using stochastic comparison to a single-server queue, for the first time. Numerical experiments demonstrate the effectiveness of the method in scenarios when the service times have heavy-tailed (Pareto), phase-type, and exponential distributions.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1386-7857
1573-7543
DOI:10.1007/s10586-025-05678-2