论文部分内容阅读
在基于主动探针的服务故障管理中,不确定性和噪声会对服务故障管理带来影响.为了降低这种影响,分析了Internet服务故障管理中存在的问题,采用二分Bayes网络建模故障和探针之间的依赖联系,二元对称信道建模噪声,并提出了不确定和噪声环境下的主动探针故障管理方案.该方案由两阶段组成:故障监测和故障诊断.在故障监测阶段,提出了在保证一定监测质量的条件下选择最小代价探针子集的GAPSA算法.在故障诊断阶段,提出了根据前一阶段发现的症状选择更多探针来获取系统详细信息的FDPSA算法;针对故障自动修复机制导致的动态性,提出了基于故障持续时间统计的假设推理算法FPTS.仿真结果证明了本文算法的有效性和效率.
In the active probe-based service fault management, the uncertainty and noise will affect the service fault management.In order to reduce the impact, this paper analyzes the problems existing in the Internet service fault management, and uses the bisection Bayes network to model the fault and Probe dependencies, binary symmetric channel modeling noise, and proactive probe fault management scheme in uncertain and noisy environments.This scheme consists of two phases: fault monitoring and fault diagnosis.In the fault monitoring phase , A GAPSA algorithm is proposed to select the minimum cost subset of probes under the condition of ensuring the quality of monitoring.At the phase of fault diagnosis, an FDPSA algorithm is proposed based on the symptoms found in the previous stage to select more probes to obtain the system detailed information. Aimed at the dynamic caused by fault auto-repair mechanism, a hypothetical inference algorithm FPTS based on fault duration statistics is proposed. The simulation results show the effectiveness and efficiency of the proposed algorithm.