Amazon AWS Certified Solutions Architect - Professional SAP-C02 Bundle

Cross-account access findings identify external principals and clear only when permissions are corrected. However, this does not establish the required outcome: approved orphaned volumes stop generating storage costs while protected volumes remain recoverable.

Tested gp3 configuration meets IO SLOs at lower monthly storage and provisioned-performance cost. However, this does not establish the required outcome: approved orphaned volumes stop generating storage costs while protected volumes remain recoverable.

Identified buckets show smaller unnecessary version and multipart storage after retention-safe cleanup. However, this does not establish the required outcome: approved orphaned volumes stop generating storage costs while protected volumes remain recoverable.

The service relies solely on application code with no rate-based edge filtering. The stated evidence instead indicates: Instance-local caches cannot share consistent ephemeral state across horizontally scaled hosts.

Every read request traverses the backing DynamoDB service despite highly reusable cacheable data. The stated evidence instead indicates: Instance-local caches cannot share consistent ephemeral state across horizontally scaled hosts.

Read-only analytical traffic saturates the production database's single writable instance. The stated evidence instead indicates: Instance-local caches cannot share consistent ephemeral state across horizontally scaled hosts.

This action targets a different technical need: An engineer repeatedly leaves management ports publicly accessible despite a configuration guardrail document. It does not replace either required solution in A or B. AWS Config checks configuration changes against evaluable rules and supports managed remediation actions; verify the two stated requirements independently rather than accepting an unrelated change.

This action targets a different technical need: A platform team learns about upcoming service maintenance only after workloads start failing. It does not replace either required solution in A or B. AWS Health can surface AWS-side events relevant to affected resources and accounts; verify the two stated requirements independently rather than accepting an unrelated change.

This action targets a different technical need: A multi-AZ production stack has never been tested with a failed Availability Zone during working hours. It does not replace either required solution in A or B. Fault injection validates real system behavior within approved experiment boundaries; verify the two stated requirements independently rather than accepting an unrelated change.

Repeated retries of one failed message prevent useful error isolation and redrive workflow. The stated evidence instead indicates: DNS simple routing cannot automatically reflect the application's health.

Imperative glue retries all upstream steps and creates duplicate side effects. The stated evidence instead indicates: DNS simple routing cannot automatically reflect the application's health.

Each team independently patches hand-built machine images without reproducible validation. The stated evidence instead indicates: DNS simple routing cannot automatically reflect the application's health.

Unattached volumes are treated as free merely because they are not mounted. The stated evidence instead indicates: Instances remain oversized based only on purchase history rather than performance evidence.

Potential savings from the same infrastructure change are double counted in separate reports. The stated evidence instead indicates: Instances remain oversized based only on purchase history rather than performance evidence.

Health checks use an unrelated always-200 endpoint that does not reflect the service readiness. The stated evidence instead indicates: Instances remain oversized based only on purchase history rather than performance evidence.

Separate point-to-point VPC peering connections lack transitive routing and are difficult to govern at scale. The specific problem here is that ordinary regional endpoints offer neither the requested global anycast entry point nor accelerator health steering.

Per-account one-off inspections do not continuously evaluate drift across accounts. The specific problem here is that ordinary regional endpoints offer neither the requested global anycast entry point nor accelerator health steering.

A broad VPC peering link was used where a service-scoped private access path was needed. The specific problem here is that ordinary regional endpoints offer neither the requested global anycast entry point nor accelerator health steering.

Events remain confined to local account buses because cross-account bus authorization is absent. The stated evidence instead indicates: All service replicas were scheduled in one zone despite multi-AZ infrastructure.

Synchronous request-to-database coupling loses or rejects work whenever the database is unavailable. The stated evidence instead indicates: All service replicas were scheduled in one zone despite multi-AZ infrastructure.

A single-Region Aurora cluster lacks a suitable crossregional disaster recovery target. The stated evidence instead indicates: All service replicas were scheduled in one zone despite multi-AZ infrastructure.

Identified buckets show smaller unnecessary version and multipart storage after retention-safe cleanup. However, this does not establish the required outcome: an incident view identifies the affected service and response latency rather than just total fleet CPU.

Compliance reports show applied baselines and chosen hosts patched only during approved windows. However, this does not establish the required outcome: an incident view identifies the affected service and response latency rather than just total fleet CPU.

A test transaction shows the slow downstream dependency and request trace span timing. However, this does not establish the required outcome: an incident view identifies the affected service and response latency rather than just total fleet CPU.

Identical input configurations yield reviewed deployment changes and detectable drift rather than ad hoc resources. However, this does not establish the required outcome: the authorized role decrypts the designated ciphertext while an unrelated role remains denied.

Cloud instances access approved services privately while disallowed API paths are rejected. However, this does not establish the required outcome: the authorized role decrypts the designated ciphertext while an unrelated role remains denied.

A rotation completes, the application authenticates with the new value, and old credentials are not exposed. However, this does not establish the required outcome: the authorized role decrypts the designated ciphertext while an unrelated role remains denied.

Source virtualization format was mistaken for a guarantee of target runtime compatibility. The stated evidence instead indicates: All assets are scheduled for direct rehost even when some are obsolete or unsuitable.

Migration scheduling ignored data classification and jurisdictional control requirements. The stated evidence instead indicates: All assets are scheduled for direct rehost even when some are obsolete or unsuitable.

A non-Kubernetes orchestrator was chosen despite strict Kubernetes extension compatibility requirements. The stated evidence instead indicates: All assets are scheduled for direct rehost even when some are obsolete or unsuitable.

Mainframe batch and transaction semantics were treated as if they were ordinary Linux workloads. The stated evidence instead indicates: Manual one-time disk copying cannot keep hundreds of moving workloads synchronized until cutover.

Migration tasks are tracked in separate team documents with no common portfolio state. The stated evidence instead indicates: Manual one-time disk copying cannot keep hundreds of moving workloads synchronized until cutover.

Traffic was switched to the container environment before dependent services and probes were verified. The stated evidence instead indicates: Manual one-time disk copying cannot keep hundreds of moving workloads synchronized until cutover.