Ruhao Xin and Juncheng Dong
We study the benchmarking platform for failure detectors. By Failure Detectors, we mean some algorithms that can determine whether a process crashes or not. By benchmarking, we mean some methods that can evaluate the performance of failure detectors based on some metrics. We propose a set of metrics to measure how well a failure detector work by analyzing heartbeat messages. Next, we design a set of language to specify general failure detectors. We then use some benchmarking algorithms to visualize the results to show what a "good" failure detector looks like. Finally, we build a software prototype that integrates every part of our work together.
Copyright © 2026. All rights reserved