Fault Tolerance

Fault tolerance starts with a failure assumption: which components may fail, and which operations must continue? This lesson compares recovery, redundancy, and checkpointing, including the state and coordination needed to make those techniques safe.

What is fault tolerance?

In a distributed application, a failed component can interrupt requests, lose data, or force expensive work to be repeated. Identify single points of failure and decide which ones must be removed to meet the service’s requirements.

Fault tolerance is the ability to continue providing specified behavior despite a defined set of component failures. State the failure types, the number of simultaneous failures, and the acceptable degraded behavior. “Fault tolerant” is incomplete without those assumptions.