What it means
It is stronger than basic redundancy because the system must detect, isolate, and work around faults without unacceptable disruption or data loss.
Why it matters
Fault-tolerant designs are appropriate only where the business impact justifies their complexity and cost. They still require testing because hidden shared dependencies can defeat the design.
Practical focus
- Define which faults the service must tolerate
- Remove shared dependencies from supposedly redundant paths
- Use graceful degradation where full operation is unnecessary
- Exercise fault scenarios under realistic load
Common mistakes
- Claiming fault tolerance without failure testing
- Replicating bad state across every copy
- Making the recovery mechanism more fragile than the service
Certification relevance
This subject appears in or supports the following certification bodies of knowledge: