Scalable distributed consensus to support MPI fault tolerance.

Conference · Sat Jan 01 00:00:00 EST 2011

DOI:https://doi.org/10.1007/978-3-642-24449-0_39· OSTI ID:1035019

Buntinas, D ^[1]

---

1. Mathematics and Computer Science

As system sizes increase, the amount of time in which an application can run without experiencing a failure decreases. Exascale applications will need to address fault tolerance. In order to support algorithm-based fault tolerance, communication libraries will need to provide fault-tolerance features to the application. One important fault-tolerance operation is distributed consensus. This is used, for example, to collectively decide on a set of failed processes. This paper describes a scalable, distributed consensus algorithm that is used to support new MPI fault-tolerance features proposed by the MPI 3 Forum's fault-tolerance working group. The algorithm was implemented and evaluated on a 4,096-core Blue Gene/P. The implementation was able to perform a full-scale distributed consensus in 305 {mu}s and scaled logarithmically.

• Research Organization: Argonne National Lab. (ANL), Argonne, IL (United States)
• Sponsoring Organization: USDOE Office of Science (SC)
• DOE Contract Number: DE-AC02-06CH11357
• OSTI ID: 1035019
• Report Number(s): ANL/MCS/CP-70156; TRN: US201204%%300
• Resource Relation: Conference: 18th EuroMPI Conference; Sep. 18, 2011 - Sep. 21, 2011; Santorini, Greece
• Country of Publication: United States
• Language: ENGLISH

Related Subjects

99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
ALGORITHMS
COMMUNICATIONS
IMPLEMENTATION
TOLERANCE