Guide · Computing
Failure domains
A guide to naming which parts of a system can fail together and which consequences cross the boundary.
Start with shared failure
Group components by what can take them down together: process, host, zone, dependency, or operator action. The useful unit is the one that changes the recovery decision.
Then trace consequences
For each boundary, record what becomes unavailable, what can continue, and what can be replayed. This turns a vague reliability concern into a reading and design checklist.
References and sources