Resilience is how a system’s latency behaves when components fail or degrade. A resilient architecture continues operating through controlled degradation instead of allowing one failure to amplify across the system. section 5.1
Separate resilience from two related architectural concerns:
- Speed measures critical-path latency for one user completing one task. section 5.1
- Scalability measures how latency changes as concurrency and data volume increase. section 5.1
- Resilience addresses failure conditions through fallback strategies, circuit breakers, graceful degradation, and redundancy. section 5.1
- Measure resilience with chaos engineering and latency monitoring during incidents rather than applying scalability fixes to failure problems. section 5.1
Cache failure
When a cache layer goes down, an unresilient system sends every request to the database and latency can rise from 50 milliseconds to 5 seconds. A resilient design falls back to a slower path and degrades toward 200 milliseconds instead. section 5.1
Treat resilience as an architectural constraint alongside functionality and maintainability. Select failure-handling patterns before incidents expose the system’s failure paths. section 5.0
















