What it means
Fault domains can exist at the hardware, rack, zone, region, network, identity, control-plane, software, or organizational level. Reliability requires understanding which failures are independent and which dependencies can create correlated loss.
Why it matters
Redundancy that shares the same hidden dependency provides less protection than it appears to. Explicit failure analysis helps teams invest where resilience materially improves service outcomes.
Practical focus
- Model component and dependency failure modes
- Define recovery time and recovery point objectives
- Use bulkheads, graceful degradation, and tested failover
- Measure reliability through service-level indicators
Common mistakes
- Counting replicas without analyzing independence
- Assuming managed services cannot fail
- Ignoring control-plane and identity dependencies
- Testing only planned maintenance rather than failure scenarios
Certification relevance
This subject appears in or supports the following certification bodies of knowledge: