Adaptive divisible load scheduling strategies for workstation clusters with unknown network resources

Debasish Ghose, Hyong Joong Kim, Tae Hoon Kim

Research output: Contribution to journalArticle

38 Citations (Scopus)

Abstract

Conventional divisible load scheduling algorithms attempt to achieve optimal partitioning of massive loads to be distributed among processors in a distributed computing system in the presence of communication delays in the network. However, these algorithms depend strongly upon the assumption of prior knowledge of network parameters and cannot handle variations or lack of information about these parameters. In this paper, we present an adaptive strategy that estimates network parameter values using a probing technique and uses them to obtain optimal load partitioning. Three algorithms, based on the same strategy, are presented in the paper, incorporating the ability to cope with unknown network parameters. Several illustrative numerical examples are given. Finally, we implement the adaptive algorithms on an actual network of processor nodes using MPI implementation and demonstrate the feasibility of the adaptive approach.

Original languageEnglish
Pages (from-to)897-907
Number of pages11
JournalIEEE Transactions on Parallel and Distributed Systems
Volume16
Issue number10
DOIs
Publication statusPublished - 2005 Oct
Externally publishedYes

Fingerprint

Divisible Loads
Scheduling
Unknown
Resources
Distributed computer systems
Adaptive algorithms
Scheduling algorithms
Partitioning
Communication
Communication Delay
Adaptive Strategies
Distributed Computing
Adaptive Algorithm
Scheduling Algorithm
Prior Knowledge
Strategy
Numerical Examples
Vertex of a graph
Estimate
Demonstrate

Keywords

  • Distributed applications
  • Divisible loads
  • Multiprocessor systems
  • Scheduling and task partitioning
  • Workstations

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Theoretical Computer Science
  • Computational Theory and Mathematics

Cite this

Adaptive divisible load scheduling strategies for workstation clusters with unknown network resources. / Ghose, Debasish; Kim, Hyong Joong; Kim, Tae Hoon.

In: IEEE Transactions on Parallel and Distributed Systems, Vol. 16, No. 10, 10.2005, p. 897-907.

Research output: Contribution to journalArticle

@article{13735bc132964717906d096b88f0d2f7,
title = "Adaptive divisible load scheduling strategies for workstation clusters with unknown network resources",
abstract = "Conventional divisible load scheduling algorithms attempt to achieve optimal partitioning of massive loads to be distributed among processors in a distributed computing system in the presence of communication delays in the network. However, these algorithms depend strongly upon the assumption of prior knowledge of network parameters and cannot handle variations or lack of information about these parameters. In this paper, we present an adaptive strategy that estimates network parameter values using a probing technique and uses them to obtain optimal load partitioning. Three algorithms, based on the same strategy, are presented in the paper, incorporating the ability to cope with unknown network parameters. Several illustrative numerical examples are given. Finally, we implement the adaptive algorithms on an actual network of processor nodes using MPI implementation and demonstrate the feasibility of the adaptive approach.",
keywords = "Distributed applications, Divisible loads, Multiprocessor systems, Scheduling and task partitioning, Workstations",
author = "Debasish Ghose and Kim, {Hyong Joong} and Kim, {Tae Hoon}",
year = "2005",
month = "10",
doi = "10.1109/TPDS.2005.117",
language = "English",
volume = "16",
pages = "897--907",
journal = "IEEE Transactions on Parallel and Distributed Systems",
issn = "1045-9219",
publisher = "IEEE Computer Society",
number = "10",

}

TY - JOUR

T1 - Adaptive divisible load scheduling strategies for workstation clusters with unknown network resources

AU - Ghose, Debasish

AU - Kim, Hyong Joong

AU - Kim, Tae Hoon

PY - 2005/10

Y1 - 2005/10

N2 - Conventional divisible load scheduling algorithms attempt to achieve optimal partitioning of massive loads to be distributed among processors in a distributed computing system in the presence of communication delays in the network. However, these algorithms depend strongly upon the assumption of prior knowledge of network parameters and cannot handle variations or lack of information about these parameters. In this paper, we present an adaptive strategy that estimates network parameter values using a probing technique and uses them to obtain optimal load partitioning. Three algorithms, based on the same strategy, are presented in the paper, incorporating the ability to cope with unknown network parameters. Several illustrative numerical examples are given. Finally, we implement the adaptive algorithms on an actual network of processor nodes using MPI implementation and demonstrate the feasibility of the adaptive approach.

AB - Conventional divisible load scheduling algorithms attempt to achieve optimal partitioning of massive loads to be distributed among processors in a distributed computing system in the presence of communication delays in the network. However, these algorithms depend strongly upon the assumption of prior knowledge of network parameters and cannot handle variations or lack of information about these parameters. In this paper, we present an adaptive strategy that estimates network parameter values using a probing technique and uses them to obtain optimal load partitioning. Three algorithms, based on the same strategy, are presented in the paper, incorporating the ability to cope with unknown network parameters. Several illustrative numerical examples are given. Finally, we implement the adaptive algorithms on an actual network of processor nodes using MPI implementation and demonstrate the feasibility of the adaptive approach.

KW - Distributed applications

KW - Divisible loads

KW - Multiprocessor systems

KW - Scheduling and task partitioning

KW - Workstations

UR - http://www.scopus.com/inward/record.url?scp=27444440372&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=27444440372&partnerID=8YFLogxK

U2 - 10.1109/TPDS.2005.117

DO - 10.1109/TPDS.2005.117

M3 - Article

AN - SCOPUS:27444440372

VL - 16

SP - 897

EP - 907

JO - IEEE Transactions on Parallel and Distributed Systems

JF - IEEE Transactions on Parallel and Distributed Systems

SN - 1045-9219

IS - 10

ER -