摘要
通过对检查点的设计,提高分布式任务管理系统的性能。介绍检查点及DMS系统构成环境,然后提出检查点实现的三种方法,即容错检查点,不进行任务迁移的抢占式调度,及任务迁移检查点。设计两个状态大小的检查点放置算法和多个状态大小的检查点放置算法,降低网络负载和检查点操作所造成的性能损失。
Performance of Business Management System Distributed（DMS） is improved by the design of Checkpoint. Introduces the Checkpoints and the DMS system, and then proposes three methods, namely, fault tolerance Checkpoint, preemptive scheduling of tasks, and task migration Checkpoint. Designs two state size checking points placement algorithm and multiple state size, reduces the performance loss caused by network load and checkpoint operation.
出处
《现代计算机》
2015年第21期3-7,共5页
Modern Computer
关键词
分布式
任务
检查点
集群
Distributed
Business
Checkpoint
Cluster
