Efficient system-level remote checkpointing technique for BLCR

Abstract

As the number of CPU cores in high-performance computing platforms continues to grow, the availability and reliability of these systems become a primary concern. As such, some solutions are physical (ie. power backup) and some are software driven. Lawrence Berkeley National Laboratory has created a system-level fault-tolerant checkpoint/restart implementation for Linux clusters. This allows processes to restart computations at the last known checkpoint if the system crashes. It also allows for manual process migration to another system when the host experiences permanent failure, assuming the checkpoint data is intact. This paper proposes an alternative method of checkpoint creation to eliminate the assumption that the checkpoint data is intact on the local disk after system failure. © 2011 IEEE.