Challenges and Issues of the Integration of RADIC into Open MPI

Authors:
Leonardo Fialho;Guna Santos;Angelo Duarte;Dolores Rexachs;Emilio Luque
Affiliations:
Computer Architecture and Operating Systems Department, University Autonoma of Barcelona., Bellaterra, Barcelona, Spain 08193;Computer Architecture and Operating Systems Department, University Autonoma of Barcelona., Bellaterra, Barcelona, Spain 08193;Departamento de Tecnologia, Universiade Estadual de Feira de Santana. Feira de Santana, Bahia, Brazil;Computer Architecture and Operating Systems Department, University Autonoma of Barcelona., Bellaterra, Barcelona, Spain 08193;Computer Architecture and Operating Systems Department, University Autonoma of Barcelona., Bellaterra, Barcelona, Spain 08193
Venue:
Proceedings of the 16th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
Year:
2009

Citing 3
Cited 0

MPICH-V2: a Fault Tolerant MPI for Volatile Nodes based on Pessimistic Sender Based Message Logging

Proceedings of the 2003 ACM/IEEE conference on Supercomputing
Fault Tolerance in Message Passing Interface Programs

International Journal of High Performance Computing Applications
Providing Non-stop Service for Message-Passing Based Parallel Applications with RADIC

Euro-Par '08 Proceedings of the 14th international Euro-Par conference on Parallel Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Parallel machines are growing in complexity and number of components which increases fault probability. Thus, MPI applications running on these machines may not reach completion. This paper presents RADIC/OMPI, which is the integration of RADIC fault tolerance architecture into Open MPI. RADIC/OMPI relies on uncoordinated checkpoints combined with pessimistic receiver-based message logs in a distributed way without the need to use any central or stable elements. Due to this, it assures the application completion automatically and transparently for users and administrators. We concluded that within certain applications RADIC/OMPI provides fault tolerance with an acceptable overhead even in the presence of consecutive faults.