A Checkpointing-Recovery Scheme for Optimistically Synchronized Parallel Computations

This paper presents a checkpointing-recovery scheme for optimistically synchronized parallel computations. The scheme relies on a checkpointing protocol, namely hybrid state saving, embedding both sparse and incremental state saving modes, and on a state recovery procedure embedding both forward and backward recovery modes. This scheme is a generalization of many previous solutions, which can be obtained as particular instances of it by selecting appropriate values for the checkpointing protocol parameters. We also present two regulating algorithms to tune the checkpointing protocol parameters on line, in order to make the protocol react to dynamic rollback behavior. The scheme is compared to previous solutions through a case study in the context of optimistically synchronized parallel discrete event simulation. The comparison has been carried out by using a classical benchmark in eight diierent conngurations. The obtained data show that our scheme allows faster execution and, in addition, keeps quite low the amount of memory used for recording state information.