Well-Architected Framework: Reliability

1. What should you identify first when designing for reliability?

B. Flows have different criticality; requirements per flow drive the design.

2. What's a reliability target expressed as a percentage of time a flow must be available?

C. Service level objectives turn business needs into measurable targets.

3. Why should a workload's SLO usually be lower than the product of its components' best SLAs suggests?

D. Calculate composite SLAs and set realistic, achievable SLOs.

4. What's failure mode analysis (FMA)?

B. FMA finds single points of failure and failure impacts early.

5. Which design removes single points of failure inside a region?

B. Multiple instances in different zones survive a datacenter failure.

6. A downstream service is failing; callers keep retrying and make it worse. Which pattern helps?

D. A circuit breaker stops calls for a while so the failing service can recover.

7. How should clients retry transient failures?

C. Backoff avoids overloading a recovering service.

8. During an outage of the recommendations service, the shop still sells products without recommendations. What's this called?

D. Non-critical features degrade while critical flows keep working.

9. Which pattern decouples a front end from a slow back end so load spikes are absorbed?

D. A queue buffers requests; the back end processes at its own pace.

10. What's the purpose of a disaster recovery plan in the reliability pillar?

A. Recovery must be planned, automated where possible, and rehearsed.

11. Which backup practice best supports reliable recovery?

C. A backup you can't restore isn't a backup.

12. What does an active-active multi-region design give compared to active-passive?

A. Active-passive is cheaper but failover takes longer.

13. What should health monitoring for reliability include?

A. A health model shows whether a flow is healthy, degraded or unhealthy.

14. What's chaos engineering?

A. Fault injection proves that the design handles failures.

15. Why should you run reliability tests in pre-production and production-like environments?

A. Load, failover and fault-injection tests build confidence.

16. What does "keep it simple" mean for reliability?

C. Simpler designs have fewer failure points and are easier to operate.

17. Why prefer managed PaaS services for reliability where possible?

C. Offloading responsibility reduces operational risk.

18. What's a health endpoint (health probe) used for?

D. Health endpoint monitoring removes failing instances from rotation.

19. What does self-preservation (self-healing) include?

D. Self-healing reduces downtime without human intervention.

20. Why should critical and non-critical flows be isolated?

A. The bulkhead pattern isolates resources per flow or consumer.