A note on uncoordinated checkpointing with finite checkpoints

摘要

This paper discusses distributed checkpointing with "Time Warp techniques", a kind of uncoordinated checkpointing technique. First, we present an evaluation model of optimal checkpoint interval for low-overhead checkpointing and recovery. Relaxing the assumption of the previous model of Soliman et al., we show a discrete time model where the number of available checkpoints that each process can keep is finite. In addition, we propose an optimal checkpointing technique, that makes a time arrangement of checkpoints more effective for a recovery point distribution, and we give numerical examples.