A fault-tolerance mechanism in grid

Grid appears as an effective technology coupling geographically distributed resources for solving large-scale problems in the wide area network. Fault tolerance in Gird system is a significant and complex issue to secure a stable and reliable performance. Until now, various technique exist for detecting and correcting faults in distributed computing sytems. Unfortunately, few energy focus on fault-tolerance in Grid environment, especially with the emergence of OGSA. A new fault-tolerant mechanism is needed to detect and recover service faults and nodes crash. Based on our previous work on Java threads state capturing and existing Mobile Agent techniques, we put forward a fault-tolerant mechanism in this paper, providing effective fault-handling and recovering methods.