Domain 3 of 4 · Chapter 4 of 5

Improving reliability

Unlock the complete study guide + 1,040 practice questions across 16 full exams.

Bundled into the existing AWS Certified Solutions Architect - Professional premium course — no separate purchase.

Included in this chapter:

  • The improvement loop: measure, assess, remediate, re-assess
  • Prove recovery with AWS FIS chaos experiments
  • Detect degradation early: Synthetics, DevOps Guru, quotas
  • Verify backups, automate recovery, and run DR drills
  • Exam-pattern recognition: which tool, which answer

Reliability-improvement tools for an existing workload (SAP-C02)

DimensionAWS FISAWS Resilience HubCloudWatch SyntheticsTrusted Advisor + Service Quotas
Primary jobInject faults to prove recovery (chaos engineering)Assess architecture vs RTO/RPO and recommend fixesSynthetic monitoring of endpoints/APIsDetect quota/limit risk before it blocks recovery
OutputExperiment results: did the workload recover?Resiliency score + alarm/SOP/test recommendationsAvailability & latency metrics, screenshots, alarms80%/100% limit checks; quota-utilization CloudWatch alarms
Reactive vs proactiveProactive (test before the real failure)Proactive (gap analysis ahead of incident)Proactive (catch outage before customers)Proactive (raise quota before it bites)
Safety guardrailStop conditions tied to CloudWatch alarms; auto rollbackRead-only assessment, no production impactRead-only probes (writes only if scripted)Read-only checks and alarms
Runs on a scheduleYes (recurring GameDays / scheduled experiments)Re-assess after each changeYes (cron/rate, as often as every minute)Trusted Advisor refreshes; alarms evaluate continuously

Decision tree

What is the reliability gap in the existing workload? Recovery unproven Not measured vs RTO/RPO Need early detection Validate it recovers from a real failure? Catch trouble before customers do? Score architecture vs RTO / RPO targets? Inject faults Backups only Probe endpoint ML anomalies Assess AWS FIS experiment (stop conditions = CloudWatch alarms) AWS Backup restore testing (restore vs RTO) CloudWatch Synthetics canary (every minute) Amazon DevOps Guru AWS Resilience Hub (resiliency score) Gaps found Quota could block scaling at peak? Yes Service Quotas alarms + Trusted Advisor (80% / 100%) Always: remove single points of failure (multi-AZ, RDS Multi-AZ / Aurora, externalized session state) and prefer automatic recovery over manual runbooks.
Cheat sheet

Cheat sheet

  • FIS injects faults; Resilience Hub scores against RTO/RPO
  • Bound every FIS experiment with a CloudWatch-alarm stop condition
  • FIS actions span instance, network, API, and whole-AZ faults
  • A Resilience Hub policy sets RTO and RPO per disruption type
  • A Resilience Hub assessment recommends alarms, SOPs, and tests
  • Synthetics canaries are scheduled scripts that run as Lambda
  • Canaries can run every minute and alarm on the synthetics namespace
  • DevOps Guru uses ML to find anomalies without static thresholds
  • DevOps Guru gives reactive and proactive insights, with SNS notify
  • Trusted Advisor flags service quotas yellow at 80%, red at 100%
  • A backup is not a recovery until you restore it
  • A restore testing plan sets schedule, resources, and recovery point
  • Restore testing turns recoverability into Backup Audit Manager evidence
  • Prefer automatic recovery over a human-run runbook
  • Drill recovery with FIS-driven GameDays so runbooks don't rot
  • Track SLIs against an SLO to make reliability measurable
  • Route 53 failover routing needs health checks. Invert to drain on purpose
  • Match the ASG grace period and warmup to instance startup time

Unlock with Premium — includes all practice exams and the complete study guide.

Also tested in

References

  1. Reliability Pillar - AWS Well-Architected Framework Well-Architected
  2. AWS Resilience Hub overview
  3. What is Amazon ElastiCache?
  4. What is Amazon DynamoDB?
  5. Amazon RDS Multi-AZ deployments with one standby
  6. High availability for Amazon Aurora
  7. What is AWS Fault Injection Service?
  8. Synthetic monitoring (canaries) - Amazon CloudWatch
  9. What is Amazon DevOps Guru?
  10. Service Quotas and Amazon CloudWatch alarms
  11. Trusted Advisor service limits checks
  12. What is AWS Health?
  13. What is AWS Backup?
  14. AWS Backup restore testing