Faulttolerance

Fault tolerance is the ability of a system to continue operating properly in the presence of faults within some of its components. It aims to provide continuity of service and data integrity by detecting faults, containing their effects, and recovering quickly. Fault-tolerant design seeks to prevent total failure rather than merely reducing the chance of failure.

Key techniques include redundancy (duplicated components, triplication, N-modular redundancy) and process diversity (N-version programming) to avoid common-mode failures; error detection and correction (parity bits, ECC memory, checksums); monitoring and heartbeat mechanisms (watchdogs); isolation and containment to prevent a fault from propagating; and graceful degradation to maintain partial functionality when full service cannot be sustained.

Recovery methods: failover to standby components, checkpointing and rollback, replay of logs, data replication across sites, erasure coding, distributed consensus protocols (Paxos, Raft) to tolerate node failures; reconfiguration and self-healing systems adapt to faults.

Applications vary: in hardware (RAID, redundant power supplies), in software (exception handling, retry logic, transactional memory), in networks (multipath routing, redundant links), and in distributed systems (replicated data stores, consensus). Metrics include availability, MTBF, MTTR, and resilience; design often trades reliability for cost, performance, and complexity.

Related concepts include high availability, reliability engineering, and resilience. Fault tolerance remains a core objective in critical systems such as avionics, data centers, and industrial control.