The Checkpoint Interval Optimization of Kernel-Level Rollback Recovery Based on the Embedded Mobile Computing System
Zhan Zhang, Decheng Zuo, Yiwei Ci, Xiaozong Yang · 2008
Due to the limited resources of the embedded mobile computing systems, such as wearable computers, PDAs or sensor nodes, reducing the overhead of the software implemented fault tolerance mechanism is a key factor in reliability design. Two checkpoint interval optimization techniques of kernel level rollback recovery mechanism are discussed. Step checkpointing algorithm modulates checkpoint intervals on-line according to the characteristcs of software or hardware environment-dependent system that the failure rate fluctuates acutely shortly after the system fails. Checkpoint size monitoring and threshold-control technique adjusts the checkpoint interval by predicting the amount of data to be saved. Combining these two techniques can effectively improve the performance of the embedded mobile computer system.