On the complexity of scheduling checkpoints for computational workflows

This paper deals with the complexity of scheduling computational workflows in the presence of Exponentially distributed failures. When such a failure occurs, rollback and recovery is used so that the execution can resume from the last checkpointed state. The goal is to minimize the expected execution time, and we have to decide in which order to execute the tasks, and whether to checkpoint or not after the completion of each given task. We show that this scheduling problem is strongly NP-complete, and propose a (polynomial-time) dynamic programming algorithm for the case where the application graph is a linear chain. These results lay the theoretical foundations of the problem, and constitute a prerequisite before discussing scheduling strategies for arbitrary DAGS of moldable tasks subject to general failure distributions.