Monte Carlo simulation on heterogeneous distributed systems: A computing framework with parallel merging and checkpointing strategies

Sorina Camarasu-Pop; Tristan Glatard; Rafael Ferreira Da Silva; Pierre Gueth; David Sarrut; Hugues Benoit-Cattin

doi:10.1016/j.future.2012.09.003

Monte Carlo simulation on heterogeneous distributed systems: A computing framework with parallel merging and checkpointing strategies

Sorina Camarasu-Pop, Tristan Glatard, Rafael Ferreira Da Silva, Pierre Gueth, David Sarrut, Hugues Benoit-Cattin

Research output: Contribution to journal › Article › peer-review

28 Scopus citations

Abstract

This paper introduces an end-to-end framework for efficient computing and merging of Monte Carlo simulations on heterogeneous distributed systems. Simulations are parallelized using a dynamic load-balancing approach and multiple parallel mergers. Checkpointing is used to improve reliability and to enable incremental results merging from partial results. A model is proposed to analyze the behavior of the proposed framework and help tune its parameters. Experimental results obtained on a production grid infrastructure show that the model fits the real makespan with a relative error of maximum 10%, that using multiple parallel mergers reduces the makespan by 40% on average, that checkpointing enables the completion of very long simulations and that it can be used without penalizing the makespan.

Original language	English
Pages (from-to)	728-738
Number of pages	11
Journal	Future Generation Computer Systems
Volume	29
Issue number	3
DOIs	https://doi.org/10.1016/j.future.2012.09.003
State	Published - Mar 2013
Externally published	Yes

Funding

This work is co-funded by the French national research agency (ANR) , hGATE project under contract no. ANR-09-COSI-004-01. It also falls into the scope of the scientific topics of the French National Grid Institute (IdG). The authors would like to thank the site administrators of the European Grid Initiative and the GGUS support for their work.

Funders	Funder number
Seventh Framework Programme	261323
Agence Nationale de la Recherche	ANR-09-COSI-004-01

Keywords

Checkpointing
Dynamic parallelization
Grid computing
Merge
Monte Carlo
Workflow

Access to Document

10.1016/j.future.2012.09.003

Cite this

@article{29ba69aba67d484a8b3a14162b970ef8,

title = "Monte Carlo simulation on heterogeneous distributed systems: A computing framework with parallel merging and checkpointing strategies",

abstract = "This paper introduces an end-to-end framework for efficient computing and merging of Monte Carlo simulations on heterogeneous distributed systems. Simulations are parallelized using a dynamic load-balancing approach and multiple parallel mergers. Checkpointing is used to improve reliability and to enable incremental results merging from partial results. A model is proposed to analyze the behavior of the proposed framework and help tune its parameters. Experimental results obtained on a production grid infrastructure show that the model fits the real makespan with a relative error of maximum 10%, that using multiple parallel mergers reduces the makespan by 40% on average, that checkpointing enables the completion of very long simulations and that it can be used without penalizing the makespan.",

keywords = "Checkpointing, Dynamic parallelization, Grid computing, Merge, Monte Carlo, Workflow",

author = "Sorina Camarasu-Pop and Tristan Glatard and {Da Silva}, {Rafael Ferreira} and Pierre Gueth and David Sarrut and Hugues Benoit-Cattin",

year = "2013",

month = mar,

doi = "10.1016/j.future.2012.09.003",

language = "English",

volume = "29",

pages = "728--738",

journal = "Future Generation Computer Systems",

issn = "0167-739X",

publisher = "Elsevier",

number = "3",

}

TY - JOUR

T1 - Monte Carlo simulation on heterogeneous distributed systems

T2 - A computing framework with parallel merging and checkpointing strategies

AU - Camarasu-Pop, Sorina

AU - Glatard, Tristan

AU - Da Silva, Rafael Ferreira

AU - Gueth, Pierre

AU - Sarrut, David

AU - Benoit-Cattin, Hugues

PY - 2013/3

Y1 - 2013/3

N2 - This paper introduces an end-to-end framework for efficient computing and merging of Monte Carlo simulations on heterogeneous distributed systems. Simulations are parallelized using a dynamic load-balancing approach and multiple parallel mergers. Checkpointing is used to improve reliability and to enable incremental results merging from partial results. A model is proposed to analyze the behavior of the proposed framework and help tune its parameters. Experimental results obtained on a production grid infrastructure show that the model fits the real makespan with a relative error of maximum 10%, that using multiple parallel mergers reduces the makespan by 40% on average, that checkpointing enables the completion of very long simulations and that it can be used without penalizing the makespan.

AB - This paper introduces an end-to-end framework for efficient computing and merging of Monte Carlo simulations on heterogeneous distributed systems. Simulations are parallelized using a dynamic load-balancing approach and multiple parallel mergers. Checkpointing is used to improve reliability and to enable incremental results merging from partial results. A model is proposed to analyze the behavior of the proposed framework and help tune its parameters. Experimental results obtained on a production grid infrastructure show that the model fits the real makespan with a relative error of maximum 10%, that using multiple parallel mergers reduces the makespan by 40% on average, that checkpointing enables the completion of very long simulations and that it can be used without penalizing the makespan.

KW - Checkpointing

KW - Dynamic parallelization

KW - Grid computing

KW - Merge

KW - Monte Carlo

KW - Workflow

UR - http://www.scopus.com/inward/record.url?scp=84870322334&partnerID=8YFLogxK

U2 - 10.1016/j.future.2012.09.003

DO - 10.1016/j.future.2012.09.003

M3 - Article

AN - SCOPUS:84870322334

SN - 0167-739X

VL - 29

SP - 728

EP - 738

JO - Future Generation Computer Systems

JF - Future Generation Computer Systems

IS - 3

ER -

Monte Carlo simulation on heterogeneous distributed systems: A computing framework with parallel merging and checkpointing strategies

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this