A Study on Adaptive Fault Detection Algorithm for Distributed Storage System

In order to establish an efficient and reliable fault detection mechanism for the distributed storage system, a mixed status piggyback adaptive fault detection algorithm was presented. In addition, DFDF, a fault detection framework for the distributed storage system was also put forward and it was able to provide load balancing and high reliability by administrating the operational mode of mutual backup and concurrent working for nodes. According to detection delay calculation based on load predictions and the adaptive adjustment of timeout parameters, a fault detection adaptive algorithm known as MSP-AFD was proposed. Experiments proved that detection performance of MSP-AFD algorithm was superior to the traditional NFD-E algorithm in various load conditions so that both performance of fault detection and self-adaptability was improved.