Pass Your Amazon AWS Certified Solutions Architect - Professional SAP-C02 Exam Easy!

KMS keys are Region-scoped unless a supported replica strategy is explicitly configured. The recovery bucket needs decryptability from destination-accessible key material.

S3 Bucket Keys can reduce some KMS calls but are not an independent decryption mechanism. Destination key policy, role permissions and actual restore tests remain necessary.

Policy JSON is not cryptographic key material and cannot decrypt a snapshot or S3 object. A real Region-local key and authorized re-encryption are needed.

Database replication does not replicate a control-plane scheduler. The secondary Region must be able to resume scheduled billing without double execution.

Traffic cutover does not reconstruct queued/scheduled jobs or ensure the downstream billing workflow is active and consistent.

Database recovery does not restore independent EventBridge rules, worker queues, secrets or job ownership. Each dependency belongs in the rehearsed DR plan.

Multi-AZ resilience helps with zone failure, not a regional disaster leaving no accessible backup copy.

Frequent same-Region snapshots can improve local RPO but do not fix the observed absence of regional recovery data or prove the one-hour RTO.

A DNS endpoint does not reconstruct a database. Recovery data and a rehearsed restore workflow must exist in the second Region.

Additional AZ coverage still shares the same regional blast radius; the business requires a separately deployed regional copy.

Cross-Region backup copies are useful but must be restored; they are not continuously updated read replica instances available for promotion.

Multi-AZ DB clusters operate within one Region and do not create the off-Region read replica needed to survive full regional failure.

A zero-capacity application environment is not functioning warm standby; it needs nonzero healthy reduced application capacity.

Drill-only capacity does not preserve a warm standby during an unannounced production failure. Running ready capacity is the defining feature.

Secondary compute alone cannot satisfy application recovery if the needed state is absent; warm standby must include tested data and dependencies.

Multi-AZ cache failover does not provide cross-Region session continuity; use explicitly replicated durable state or accept a safe reauthentication flow.

A parameter group defines configuration, not runtime customer session data, so payment continuity still fails.

Snapshots are not synchronous session sharing or continuity, and relying solely on them may lose recent transient session information.

A second copy may improve resilience but does not demonstrate actual restoration viability or measured recovery time.

Audit reporting can evaluate controls but cannot by itself prove a successful actual RDS, EBS or EFS restoration within the target.

A completed backup copy is not the same as a timed end-to-end restore, data validation and application readiness.

A healthy generic endpoint is already proven misleading. More frequent polling of the wrong signal does not detect failed settlements.

Route 53 alarm-backed checks support documented alarm types and standard-resolution metrics; high-resolution or metric-math alarms are not universally supported.

Latency routing and health are separate; it can continue selecting a fast but functionally broken Region without the business-failure signal.

Documented objectives do not demonstrate measured recovery, missing dependencies or the age of actual restored committed data.

The end-to-end recovery target includes application readiness and routing; database-only timing underreports actual outage duration.

Infrastructure timing omits data loss, replay lag, application validation, external dependencies and client routing; a full rehearsal is needed.

Capacity adjustments in an unreachable Region cannot redirect failing client requests.

Global tables replicate data, but client SDK requests continue to target their configured Region unless the application deliberately changes routing.

A DNS name mapped to the failed service Region will not ensure clients invoke the healthy regional table endpoint.

An ordinary live replication rule primarily handles new eligible changes; already existing objects require supported batch replication/backfill.

Inventory identifies objects and can supply manifests, but does not itself replicate historic object content without a Batch Replication job.

Deleting production data for backfill is destructive and not necessary to replicate existing objects.