2002 · 74 citations · 7 references
ComputingComputer EngineeringComputer ScienceFault ToleranceFault-tolerant ComputingParallel ComputingFault Tolerance DesignFault Tolerance In CommunicationFault Tolerance In Parallel SystemsFault-tolerant CommunicationReliable Distributed SystemsReliable System DesignEvent HandlingFault CoverageFault Tolerance ArchitectureFault Tolerance In MultiprocessorsFault-tolerant Computation ModelsFault-tolerant MiddlewareMessage Passing MiddlewareRecovery Management
MPI has proven effective for parallel applications in situations with neither QoS nor fault handling. Emerging environments motivate fault-tolerant MPI middleware. Environments include space-based, wide-area/web/meta computing and scalable clusters. MPI/FT, the system described in the paper, trades off sufficient MPI fault coverage against acceptable parallel performance, based on mission requirements and constraints. MPI codes are evolved to use MPI/FT features. Non-portable code for event handlers and recovery management is isolated. User-coordinated recovery, checkpointing, transparency and event handling, as well as evolvability of legacy MPI codes form key design criteria. Parallel self-checking threads address four levels of MPI implementation robustness, three of which are portable to any multithreaded MPI. A taxonomy of application types provides six initial fault-relevant models; user-transparent parallel nMR computation is thereby considered. Key concepts from MPI/RT-real-time MPI-are also incorporated into MPI/FT, with further overt support for MPI/RT and MPI/FT in applications possible in future.
7
Richard D. Schlichting, Fred B. Schneider · ACM Transactions on Computer Systems · 1983 · 769 citations · Full text
CoCheck: checkpointing and process migration for MPI
Georg Stellner · 2002 · 339 citations
Mark Hayden · eCommons (Cornell University) · 1998 · 309 citations · Full text