摘要： 采用主动探测的方法对分布式应用进行故障管理。针对分布式应用的特点，在主动探测的故障检测和故障诊断阶段中，分别提出基于成本效益平衡的检测集选择算法和诊断集中考虑探测成本的最小贪婪搜索算法。实验结果表明，算法在探针数量、探测时间、探测流量上都有明显的改进。

关键词: 主动探测, 贝叶斯网, 分布式应用, 故障诊断

Abstract: This paper uses active prode for fault management of distributed applications. According to the features of distributed applications, probe set selection algorithm based on cost-benefit balance is proposed in the fault detection phase, and diagnosis probe selection algorithm based on minimum greedy search and probe cost is proposed in the fault diagnosis phase. Experimental results show that the number of probes, detection time and detection traffic have significant improvement.

Key words: active probe, Bayesian network, distributed applications, fault diagnosis

中图分类号: 

• TP393