基于动态备份的容错网格任务调度*

提出了一种基于自适应备份的网格容错任务调度算法:最高百分之k备份算法.该算法对任务的安全需求和资源的信任等级进行匹配,在系统安全等级较低并且网络和主机可能失效的网格环境中进行容错任务调度.调度时,该算法根据整个网格系统的安全状况,对具有最高安全需求的百分之k的任务进行动态备份,任务备份数根据系统安全状况自适应变化,并对失败的任务重新调度.仿真结果表明,该算法可以有效提高不安全网格环境下的任务调度成功率,具有很好的容错性和可扩展性,优于固定备份数的网格任务调度算法.

Fault tolerant grid job scheduling based on dynamic replication

A fault tolerant scheduling algorithm based on dynamic replication was proposed, namely, k percent replication (KPR). KPR scheduled the jobs by matching the user security demand and resource trust level. Only the k percent jobs with highest security demands would be replicated and the number of the job replications changed adaptively according to the dynamics of grid security. In risky and failure prone grids, KPR is fault tolerant due to its dynamic replication and rescheduling mechanism. Simulation results show that KPR is fault tolerant, scalable, and better than non security aware and fixed number replication based algorithms.