失效检测是分布式系统特别是云平台容错的关键组成部分。然而由于网络状态的不断变化,要实现对失效的快速准确的检测变得比较困难。为应对这种情况,本文提出了一种基于Voherra级数的高效的自适应失效检测机制。该机制基于Volterra滤波器实现,能够只利用很少的训练样本就可以对时间序列做出很好的预测,且可根据预测误差来自适应地调整以使得预测结果更准确。为了适应云平台中各功能模块对失效检测的不同QoS需求.本文提出的失效检测机制在统计反馈部分引入调节因子仅,可以方便地满足不同应用的QoS需求。
Failure detector is one of the key components of distributed system especially for cloud computing platform. However, it becomes much more difficult to achieve rapid and accurate detection due to the changing network status. To deal with this situation, we proposed an adaptive and efficient mechanism for failure detection based on volterra series. The mechanism is based on volterra filter, can make a good time series prediction by using only a few training samples, and can be self-adapted according to the prediction error to make the predictions more accurate. Moreover,the failure detection mechanism proposed can easily meet the QoS requirements of different users by in introducing of the feedback regulatory factor α in the statistical part.