Skip to content

Partial Failure

Unlike local systems where everything works or nothing works, distributed systems experience partial failure - some components fail while others continue, creating indeterminate states that are harder to reason about than total failure.

Tension Addressed

Simplicity vs. Fault Tolerance

Recognizable As

Timeout handling without knowing if operation succeeded; idempotency tokens; health checks per service; circuit breakers; bulkheads; the "gray failure" problem

Canonical Sources

Waldo et al., "A Note on Distributed Computing" (1994); Huang et al., "Gray Failure: The Achilles' Heel of Cloud-Scale Systems" (2017)

Transfers To

Any distributed system, error handling design, resilience engineering, chaos engineering

This concept explains why distributed systems require fundamentally different error handling than local systems. You cannot simply treat remote calls like local calls - the partial failure modes create uncertainty that local calls never have.