FCKPT:A fine-grained incremental checkpoint for PCM
Xu Li, Kai Lü · 2011
System failures will become more frequent for future large scale systems and issues of fault-tolerance will be tremendously important. Checkpoint-restart techniques are very important fault-tolerance approaches for these systems. As we know, the overhead of checkpoint will influence the system performance seriously. Recently, the emerging Phase Change Memory technology is becoming available and can replace disks with the fast accesses and byte-address features. However, previous research has demonstrated that using conventional incremental checkpointing techniques directly on PCM doesn't perform well. In this paper, we propose and implement a new incremental checkpoint scheme named FCKPT, which can detect exact modified bytes of a dirty page. Our experiment results on several well-known benchmarks show that FCKPT can reduce checkpoint file size from 20%-50% without other overhead when compared to conventional page-level incremental checkpoint on PCM.