Fault tolerance and failure isolation

A system is said to be fault tolerant when it is capable of operating even if some of the pieces fail or malfunction. Typically, fault tolerance is a matter of degree: where the level of sub-component failure is either countered by other parts of the system or the degradation is gradual rather than an absolute shutdown. Faults can occur on many levels: software, hardware, or networking. A fault tolerant piece of software needs to continue to function in the face of a partial outage along any of these layers.

In a blockchain, fault tolerance on the individual hardware level is handled by the existence of multiple duplicate computers for every function—the miners in bitcoin or proof of work systems or the validators in PoS and related systems. If a computer has a hardware fault, then either it will not validly sign transactions in consensus with the network or it will simply cease to act as a network node—the others will take up the slack.