What it means
Single points of failure can be technical, organizational, geographic, procedural, or contractual. Examples include one power feed, one identity provider, one administrator with essential knowledge, one network route, one cloud region, one supplier, or one untested recovery credential.
How it works
Resilience work identifies critical dependencies, evaluates common-cause failure, and adds redundancy, diversity, failover, cross-training, spares, alternate procedures, and tested recovery. Redundancy is useful only when the alternatives do not fail for the same reason.
Two application servers do not provide meaningful resilience if both depend on one database, one storage array, one certificate, and one administrator account. The dependency chain still contains several single points of failure.
Practical focus
- Map end-to-end dependencies for critical services
- Look for common power, network, identity, supplier, and administrative dependencies
- Test failover under realistic conditions
- Cross-train essential roles and protect emergency access
Common mistakes
- Counting duplicate components without checking shared dependencies
- Ignoring people and procedures
- Buying redundancy but never exercising it
Certification relevance
This subject appears in or supports the following certification bodies of knowledge: