Amazon AWS Certified Solutions Architect - Professional SAP-C02 Bundle

Terminating all instances in one zone leaves healthy targets and Auto Scaling replaces lost capacity. However, this does not establish the required outcome: excess requests return expected throttling responses while permitted traffic continues under peak simulation.

New DNS lookups return the secondary endpoint after a primary health check changes to unhealthy. However, this does not establish the required outcome: excess requests return expected throttling responses while permitted traffic continues under peak simulation.

Hot session lookups avoid repeated database queries and failure recovery preserves acceptable service behavior. However, this does not establish the required outcome: excess requests return expected throttling responses while permitted traffic continues under peak simulation.

Metrics and logs are not organized into actionable per-service operational signals. The stated evidence instead indicates: Uncorrelated logs hide where the end-to-end latency accumulates.

Service-specific Health notifications are not routed into incident processes. The stated evidence instead indicates: Uncorrelated logs hide where the end-to-end latency accumulates.

Health checks use an unrelated always-200 endpoint that does not reflect the service readiness. The stated evidence instead indicates: Uncorrelated logs hide where the end-to-end latency accumulates.

Events remain confined to local account buses because cross-account bus authorization is absent. The specific problem here is that individual Security Hub instances were not centrally delegated or aggregated.

The remote principal has S3 object access but lacks required key-policy and IAM decrypt permission. The specific problem here is that individual Security Hub instances were not centrally delegated or aggregated.

Public ACM-issued certificates were wrongly assumed to provide a corporate private CA chain. The specific problem here is that individual Security Hub instances were not centrally delegated or aggregated.

Single-AZ database storage and compute are a single zonal failure domain. The stated evidence instead indicates: Repeated retries of one failed message prevent useful error isolation and redrive workflow.

Every reporting query was sent to the cluster writer endpoint despite available readers. The stated evidence instead indicates: Repeated retries of one failed message prevent useful error isolation and redrive workflow.

Read-only analytical traffic saturates the production database's single writable instance. The stated evidence instead indicates: Repeated retries of one failed message prevent useful error isolation and redrive workflow.

Events remain confined to local account buses because cross-account bus authorization is absent. The stated evidence instead indicates: There is no independently functioning VPN path when private connectivity goes down.

Separate point-to-point VPC peering connections lack transitive routing and are difficult to govern at scale. The stated evidence instead indicates: There is no independently functioning VPN path when private connectivity goes down.

The inspection design allows forward and return packets to use different stateful inspection appliances. The stated evidence instead indicates: There is no independently functioning VPN path when private connectivity goes down.

A single-Region Aurora cluster lacks a suitable crossregional disaster recovery target. The stated evidence instead indicates: No archive or retention mechanism was configured before the subscriber outage.

A broad VPC peering link was used where a service-scoped private access path was needed. The stated evidence instead indicates: No archive or retention mechanism was configured before the subscriber outage.

Backup images exist but no continuously replicated recovery environment or launch plan is configured. The stated evidence instead indicates: No archive or retention mechanism was configured before the subscriber outage.

This action targets a different technical need: Financial backups must survive destruction of the primary application account by a compromised administrator. It does not replace either required solution in A or B. Cross-account backup isolation reduces the effect of compromise of an application account; verify the two stated requirements independently rather than accepting an unrelated change.

This action targets a different technical need: Dozens of overprovisioned EC2 instances need evidence-based CPU and memory sizing recommendations before changing families. It does not replace either required solution in A or B. Compute Optimizer uses resource metrics and recommendations but must be checked against workload SLOs; verify the two stated requirements independently rather than accepting an unrelated change.

This action targets a different technical need: Governance wants one AWS Config compliance view from many accounts and Regions. It does not replace either required solution in A or B. AWS Config aggregators consolidate configuration and compliance data without changing source resources; verify the two stated requirements independently rather than accepting an unrelated change.

Actual eligible usage receives commitment coverage while variable spikes remain flexible. However, this does not establish the required outcome: worker count falls during light load and rises when queue age approaches the delivery deadline.

Zonal failover promotes a standby and the application reconnects after the expected interruption. However, this does not establish the required outcome: worker count falls during light load and rises when queue age approaches the delivery deadline.

A simulated interruption causes bounded rework and lowers total compute cost without losing final results. However, this does not establish the required outcome: worker count falls during light load and rises when queue age approaches the delivery deadline.

An administrator's disable request is denied while an authorized central management path remains available. However, this does not establish the required outcome: a synthetic endpoint failure directs new DNS resolutions to the secondary healthy stack within the planned TTL behavior.

Newly created and eligible changed objects arrive in the destination bucket with expected encryption. However, this does not establish the required outcome: a synthetic endpoint failure directs new DNS resolutions to the secondary healthy stack within the planned TTL behavior.

Replica lag and promotion tests meet defined RPO and RTO targets during planned and simulated events. However, this does not establish the required outcome: a synthetic endpoint failure directs new DNS resolutions to the secondary healthy stack within the planned TTL behavior.

This action targets a different technical need: An internet service must survive one Availability Zone failure while adjusting to fluctuating weekday usage. It does not replace either required solution in A or B. Multi-AZ target placement and health-driven scaling provide elastic zonal resilience; verify the two stated requirements independently rather than accepting an unrelated change.

This action targets a different technical need: A service currently stores a database password in deployment environment text and needs managed rotation without code updates. It does not replace either required solution in A or B. Secrets Manager can store and rotate secrets with applications retrieving current authorized values; verify the two stated requirements independently rather than accepting an unrelated change.

This action targets a different technical need: A worldwide public application targets near-zero regional failover time and already tolerates multi-Region data conflict semantics. It does not replace either required solution in A or B. Active-active architectures provide immediate regional capacity but require safe multi-Region data behavior; verify the two stated requirements independently rather than accepting an unrelated change.

Small traffic shares move to the new stack first; an error alarm redirects traffic to the proven stack. However, this does not establish the required outcome: writer CPU usage decreases while lag-tolerant reporting requests use replica endpoints.

Records are ingested within supported throughput limits without hot-shard throttling under replayed load. However, this does not establish the required outcome: writer CPU usage decreases while lag-tolerant reporting requests use replica endpoints.

Clients use accelerator anycast addresses and connect to the healthy regional endpoint with improved path latency. However, this does not establish the required outcome: writer CPU usage decreases while lag-tolerant reporting requests use replica endpoints.