Services, systems, and/or databases eventually provide back pressure when they fail or get overloaded. The idea is to design in back pressure to let the system degrade gracefully rather than fail chaotically.