Common Mechanisms for Supporting Fault Tolerance in DSM and Message Passing Systems
Ramamurthy Badrinath, Christine Morin, Centre National de la Recherche Scientifique (CNRS), 35 - Rennes (France). Inst . de Recherche en Informatique et Systemes Aleatoires (IRISA), Rennes-1 Univ., 35 (France). Inst. de Recherche en Informatique et Systemes Aleatoires (IRISA), Institut National des Sciences Appliquees de Rennes (INSA), 35 (France). Inst . de Recherche en Informatique et Systemes Aleatoires (IRISA), Institut National de Recherche en Informatique et en Automatique (INRIA), 35 - Rennes (France). Inst. de Recherche en Informatique et Systemes Aleatoires (IRISA) · OpenGrey (Institut de l'Information Scientifique et Technique) · 2002
Backward error recovery involving checkpointing and restart of tasks is an important component of any system providing fault tolerance to applicati- ons distributed over a network. A central problem to checkpointing and recovery is the ability to track dependencies and arrive at a consistent global checkpoint. Traditionally literature treats one of either distributed shared memory (DSM) or message passing as the interprocess communication mechanism when considering the issue of fault tolerance. This paper describes preliminary investigation into common mechanisms that can be implemented to support a wide variety of protocols in both shared memory and message passing systems. In effect it can be used in a system that combines both these IPC mechanisms.