The system should continue to:
Since no single component can guarantee 100% uptime (even the most expensive hardware eventually fails), the architecture must be design where individual components can fail without affecting the availability of the entire system.