Resilient systems keep working even when there is some error. Well, not any error. That would be a miracle, or simply nonsense! An error generally is an error. If the Armageddon comes and it is the end of the world as we know it, even resilient systems will not be responsive. For smaller disruptions, however, there may be some cure to make the systems resilient.

There are techniques that may help if only a disk fails, there is a power outage, or there is a programming error. Systems may be replicated, so when one of the instances stops responding, some other instance may take up the task of the failing one and can go on working. Systems prone to errors may be isolated from each other in terms of space or time. When there is an earthquake or flood at one location, the other location may still go on working. If different components do not need to communicate in real time and messages are stored and forwarded in a reliable manner, then this is not a problem even if the two systems are never available at the same time. They can still cooperate taking up the messages, performing the task they are supposed to, and sending out the resulting message afterwards.

Errors in the system have to be addressed even if the system remains responsive. Errors do not affect the responsiveness of a resilient system, but the level of resilience decreases and should be restored.