Abstract: Promptly reacting to failures is key for highly available datacenter services. At the core, failure detectors (FDs) have to rely on timeouts which are hard to set correctly on top of heavily ...