An Adaptive Randomized fault detection service

Fault detection is a fundamental element for management, replication, load balancing, and other services of large-scale distributed systems. But the fault detection services in such systems scale badly in the number of members that are being monitored. This paper describes a new protocol named Adaptive Randomized Gossip-based Fault Detection service (ARGoFD) which based on gossiping that does scale well and provides timely detection. We gave out a detailed description of the protocol and analyzed its main algorithms. At last we did the comparison experiments. The experiment result shows that ARGoFD could effectively reduce the redundant messages amount and increase the detection convergence rate.