A Failure Detection Model Based on Message Delay Prediction

Abstract

Failure detection is a key technology to implement a high reliable system. It is usually based on overtime mechanism to determine whether a process is failure or not. With the development of network, old failure detectors without adaptive mechanism can not meet the requirements of QoS of application all the time. Adaptive failure detection requires that the failure detectors can dynamically adjust the detecting quality according to the requirements of applications and the variations of network. A new failure detection model based on the predicted message delay is proposed in this paper. An adaptive failure detection algorithm is discussed and realized, which is based on the prediction from historical messages delay time. Experimental results show that the algorithm can satisfy the user’s demand of QoS on the failure detector to some extent.