Abstract:

A new self-adaptable failure detection algorithm is proposed in this paper based on the premise that the delay and loss of the heartbeat information is a random probability. This algorithm can realize the random accuracy requirement and adapt itself to the change of network environment. The test result shows that this algorithm can effectively decrease the error and detecting time will not grow because of the decrease of error. Therefore, it is very effective.

Key words: distributed system, faulttolerance, failure detection, adaptive failure detection algorithm