利用分布式哈希表构建带副本的检查点

利用分布式哈希表构建了一种带副本的检查点文件共享机制,可以使因特网上的计算资源共享平台由于节点异常退出时恢复或重新启动任务的执行.所有关于该检查点文件的操作都是在非集中式的分布式哈希表中完成,可以很好的跟踪、共享、回收在存储点上的检查点文件,很少需要后台服务器的参入与合作.把检查点保存在其相应的存储点上,通过哈希表中的标准函数来访问指向存储点的指针,最终完成检查点文件的基本操作.仿真结果表明:本文的检查点机制比传统的检查点策略具有比较快的任务通信轮回时间,同时计算平台也可以获得比较好的任务吞吐量,减少通信的负载.

Creating a Replicating Checkpoint Using a DHT-based P2P Networks

A replicating checkpoint mechanism for large scale computing resource sharing platform over the Internet was presented and constructed.With this technique any node in the computing platform can resume a failed task provided that the checkpoint file is available in the P2P infrastructure.Almost all interactions needed to replicate checkpoints are decentralized among the DHT(Distributed Hash Table),thus containing the load on the central supervisor.The DHT does not hold the checkpoints,but only pointers to th...