Resilience Architecture — AWS Advanced Networking (ANS-C01)
Recovery target wording signals which layer to fix
Scenarios use "no single point of failure," "99.99% availability," and "RTO under 15 minutes" as distinct signals pointing to different architectural layers. No single point of failure targets load balancer and compute tiers. Sub-15-minute RTO targets DNS failover and data replication. An Aurora Global Database satisfies near-zero RPO across regions; Route 53 health-check failover handles the traffic switch. Answers that address one layer while ignoring the other will not satisfy the full constraint.
What This Pattern Tests
The exam gives availability requirements and tests whether you design the right resilience tier. Multi-AZ deployments (RDS Multi-AZ, ECS across AZs, ALB cross-zone) protect against single AZ failure — sufficient for 99.9% to 99.99% SLAs. Multi-Region with Route 53 failover protects against regional failures — needed for 99.999% SLAs. Cell-based architecture with shuffle sharding limits blast radius for individual customer failures. The trap is designing multi-region for a 99.9% SLA (over-provisioning) or single-AZ for a 99.99% SLA (under-provisioning). Aurora Global Database replicates across regions with <1s lag — but only needed when the SLA demands regional failover.
Decision Axis
SLA target maps to resilience tier. 99.9% = Multi-AZ. 99.99% = Multi-AZ with auto-scaling. 99.999% = Multi-Region active-active.
Associated Traps
More Top Traps on This Exam
Decision Rules
Whether edge-layer origin-group failover (CloudFront, HTTP-error-code triggered, no DNS TTL dependency) or DNS-layer health-check failover (Route 53, TTL-bounded) satisfies a seconds-level RTO requirement for cacheable HTTP traffic served worldwide.
Choose CloudFront origin group failover over Route 53 DNS health-check failover when the recovery constraint is sub-second, HTTP-error-code-triggered rerouting for cacheable content served from a global edge layer.
Determine whether the edge failover mechanism introduces a caching layer whose TTL creates a post-failover stale-data window, causing the architecture to satisfy RTO while violating RPO=0, and select the mechanism that eliminates the cache layer entirely.
Whether CloudFront origin-group failover (HTTP-error-code triggered at the edge, sub-10-second recovery) or Route 53 failover routing (TTL plus health-check-interval dependent, 30-300 second recovery) satisfies the stated RTO for cacheable HTTP content served globally.
Domain Coverage
Difficulty Breakdown