Fault tolerance of MPI applications in exascale systems: The ULFM solution

Nuria Losada; Patricia González; María J. Martín; George Bosilca; Aurélien Bouteiller; Keita Teranishi

doi:10.1016/j.future.2020.01.026

Fault tolerance of MPI applications in exascale systems: The ULFM solution

Nuria Losada
, Patricia González
, María J. Martín
, George Bosilca
, Aurélien Bouteiller
, Keita Teranishi

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

51 Scopus citations

Abstract

The growth in the number of computational resources used by high-performance computing (HPC) systems leads to an increase in failure rates. Fault-tolerant techniques will become essential for long-running applications executing in future exascale systems, not only to ensure the completion of their execution in these systems but also to improve their energy consumption. Although the Message Passing Interface (MPI) is the most popular programming model for distributed-memory HPC systems, as of now, it does not provide any fault-tolerant construct for users to handle failures. Thus, the recovery procedure is postponed until the application is aborted and re-spawned. The proposal of the User Level Failure Mitigation (ULFM) interface in the MPI forum provides new opportunities in this field, enabling the implementation of resilient MPI applications, system runtimes, and programming language constructs able to detect and react to failures without aborting their execution. This paper presents a global overview of the resilience interfaces provided by the ULFM specification, covers archetypal usage patterns and building blocks, and surveys the wide variety of application-driven solutions that have exploited them in recent years. The large and varied number of approaches in the literature proves that ULFM provides the necessary flexibility to implement efficient fault-tolerant MPI applications. All the proposed solutions are based on application-driven recovery mechanisms, which allows reducing the overhead and obtaining the required level of efficiency needed in the future exascale platforms.

Original language	English
Pages (from-to)	467-481
Number of pages	15
Journal	Future Generation Computer Systems
Volume	106
DOIs	https://doi.org/10.1016/j.future.2020.01.026
State	Published - May 2020
Externally published	Yes

Funding

This work was supported by the Ministry of Economy and Competitiveness of Spain and FEDER funds of the EU (Project TIN2016-75845-P) and by the Galician Government (Xunta de Galicia, Spain) under the Consolidation Program of Competitive Research (ref. ED431C 2017/04). This research was also supported by the National Science Foundation of the United States under award NSF-SI2 #1664142, and by the Exascale Computing Project, United States (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration, United States. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC. a wholly owned subsidiary of Honeywell International, Inc. for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-NA0003525. This work was supported by the Ministry of Economy and Competitiveness of Spain and FEDER funds of the EU (Project TIN2016-75845-P) and by the Galician Government ( Xunta de Galicia, Spain ) under the Consolidation Program of Competitive Research (ref. ED431C 2017/04 ). This research was also supported by the National Science Foundation of the United States under award NSF-SI2 #1664142 , and by the Exascale Computing Project, United States ( 17-SC-20-SC ), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration, United States . Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Keywords

Application-level checkpointing
MPI
Resilience
ULFM

Access to Document

10.1016/j.future.2020.01.026

Cite this

@article{7674d5b006ee4311be8e967ff166e1bb,

title = "Fault tolerance of MPI applications in exascale systems: The ULFM solution",

abstract = "The growth in the number of computational resources used by high-performance computing (HPC) systems leads to an increase in failure rates. Fault-tolerant techniques will become essential for long-running applications executing in future exascale systems, not only to ensure the completion of their execution in these systems but also to improve their energy consumption. Although the Message Passing Interface (MPI) is the most popular programming model for distributed-memory HPC systems, as of now, it does not provide any fault-tolerant construct for users to handle failures. Thus, the recovery procedure is postponed until the application is aborted and re-spawned. The proposal of the User Level Failure Mitigation (ULFM) interface in the MPI forum provides new opportunities in this field, enabling the implementation of resilient MPI applications, system runtimes, and programming language constructs able to detect and react to failures without aborting their execution. This paper presents a global overview of the resilience interfaces provided by the ULFM specification, covers archetypal usage patterns and building blocks, and surveys the wide variety of application-driven solutions that have exploited them in recent years. The large and varied number of approaches in the literature proves that ULFM provides the necessary flexibility to implement efficient fault-tolerant MPI applications. All the proposed solutions are based on application-driven recovery mechanisms, which allows reducing the overhead and obtaining the required level of efficiency needed in the future exascale platforms.",

keywords = "Application-level checkpointing, MPI, Resilience, ULFM",

author = "Nuria Losada and Patricia Gonz{\'a}lez and Mart{\'i}n, \{Mar{\'i}a J.\} and George Bosilca and Aur{\'e}lien Bouteiller and Keita Teranishi",

note = "Publisher Copyright: {\textcopyright} 2020",

year = "2020",

month = may,

doi = "10.1016/j.future.2020.01.026",

language = "English",

volume = "106",

pages = "467--481",

journal = "Future Generation Computer Systems",

issn = "0167-739X",

publisher = "Elsevier",

}

TY - JOUR

T1 - Fault tolerance of MPI applications in exascale systems

T2 - The ULFM solution

AU - Losada, Nuria

AU - González, Patricia

AU - Martín, María J.

AU - Bosilca, George

AU - Bouteiller, Aurélien

AU - Teranishi, Keita

PY - 2020/5

Y1 - 2020/5

N2 - The growth in the number of computational resources used by high-performance computing (HPC) systems leads to an increase in failure rates. Fault-tolerant techniques will become essential for long-running applications executing in future exascale systems, not only to ensure the completion of their execution in these systems but also to improve their energy consumption. Although the Message Passing Interface (MPI) is the most popular programming model for distributed-memory HPC systems, as of now, it does not provide any fault-tolerant construct for users to handle failures. Thus, the recovery procedure is postponed until the application is aborted and re-spawned. The proposal of the User Level Failure Mitigation (ULFM) interface in the MPI forum provides new opportunities in this field, enabling the implementation of resilient MPI applications, system runtimes, and programming language constructs able to detect and react to failures without aborting their execution. This paper presents a global overview of the resilience interfaces provided by the ULFM specification, covers archetypal usage patterns and building blocks, and surveys the wide variety of application-driven solutions that have exploited them in recent years. The large and varied number of approaches in the literature proves that ULFM provides the necessary flexibility to implement efficient fault-tolerant MPI applications. All the proposed solutions are based on application-driven recovery mechanisms, which allows reducing the overhead and obtaining the required level of efficiency needed in the future exascale platforms.

AB - The growth in the number of computational resources used by high-performance computing (HPC) systems leads to an increase in failure rates. Fault-tolerant techniques will become essential for long-running applications executing in future exascale systems, not only to ensure the completion of their execution in these systems but also to improve their energy consumption. Although the Message Passing Interface (MPI) is the most popular programming model for distributed-memory HPC systems, as of now, it does not provide any fault-tolerant construct for users to handle failures. Thus, the recovery procedure is postponed until the application is aborted and re-spawned. The proposal of the User Level Failure Mitigation (ULFM) interface in the MPI forum provides new opportunities in this field, enabling the implementation of resilient MPI applications, system runtimes, and programming language constructs able to detect and react to failures without aborting their execution. This paper presents a global overview of the resilience interfaces provided by the ULFM specification, covers archetypal usage patterns and building blocks, and surveys the wide variety of application-driven solutions that have exploited them in recent years. The large and varied number of approaches in the literature proves that ULFM provides the necessary flexibility to implement efficient fault-tolerant MPI applications. All the proposed solutions are based on application-driven recovery mechanisms, which allows reducing the overhead and obtaining the required level of efficiency needed in the future exascale platforms.

KW - Application-level checkpointing

KW - MPI

KW - Resilience

KW - ULFM

UR - https://www.scopus.com/pages/publications/85078202267

U2 - 10.1016/j.future.2020.01.026

DO - 10.1016/j.future.2020.01.026

M3 - Article

AN - SCOPUS:85078202267

SN - 0167-739X

VL - 106

SP - 467

EP - 481

JO - Future Generation Computer Systems

JF - Future Generation Computer Systems

ER -

Fault tolerance of MPI applications in exascale systems: The ULFM solution

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this