复杂失效分布下的动态检查点设置

检查点技术是保障计算机系统可靠性的一种常用方法.通常假定系统失效的发生服从泊松分布,因此检查点是等间隔设置的.但近几年公布的现场数据表明以上假定不合实际.因此,首先利用现场失效数据对固定检查点间距方法的适应性进行分析,并提出两种检查点动态设置方法,它们根据系统前个阶段的失效信息动态地设置下一个检查点.模拟实验结果表明,所提出的方法在复杂失效分布下所获得的效果达到或者优于最优固定检查点的水平.

Dynamic Checkpoint Placement Algorithms Under Complex Failure Distribution

Checkpointing is a commonly used technique that protects computer system against failures.Former researches often assumed the failure distribution to be Poisson distribution,and thus derived a fixed interval between checkpoints for the optimal solution.But recently published field failure data shows that the above assumption is not realistic.This paper initially studied the applicability of the fixed interval checkpoint method using the field failure data.Then two dynamic checkpoint placement methods are pr...