An enhanced model-based checkpointing protocol

Checkpointing and rollback recovery are widely used techniques to handle failures in distributed computing systems. Usually we avoid taking checkpoints that are useless during the recovery process. Communication-Induced checkpointing algorithms guarantee the usefulness of all the checkpoints and provide considerable autonomy with relatively low overhead. In this paper, we propose an enhanced Communication-Induced checkpointing algorithm. Our algorithm is likely to have less checkpointing overhead than an existing algorithm in the literature.