Amazon AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 38

Question 370

A payment API can commit before its response is lost. A client may retry after a timeout. Which protocol-level design prevents duplicate payment side effects?

Generate a new payment identifier for each retry.
Use stable payment idempotency keys supported by the provider, reconcile the first attempt and only retry under that same operation identity.
Assume any timeout means no funds changed.
Treat SQS FIFO message deduplication as the sole safeguard for payment provider commits occurring after queue delivery retries outside the deduplication window.

Explanation

Correct Answer

Timeout creates uncertainty, not definite failure; authoritative status and deduplication protect money-moving transactions.

Incorrect Answers

Separate identifiers may create separate charges.

An acknowledgment can be lost after commit.

Queue deduplication does not guarantee exactly-once financial side effects across provider timeouts or extended retries. The payment transaction needs a stable idempotency identity.

Question 371

A disaster-recovery design review says the business can lose at most 15 minutes of data and must restore service within two hours. What should guide architecture selection?

Define RPO as application restart time and RTO as historical data-retention duration.
Treat 15 minutes as RPO and two hours as RTO; choose and test replication and restore controls against both thresholds.
Choose the least expensive archival storage without testing either constraint.
Measure only average HTTP latency and assume it equals both disaster-recovery objectives.

Explanation

Correct Answer

RPO limits acceptable data loss, while RTO limits outage duration; each needs separately demonstrated evidence.

Incorrect Answers

Those definitions swap recovery objectives.

Price cannot replace business recovery requirements.

Normal request latency is not recovered data age or elapsed restoration time.

Question 372

A team creates daily backups of a database but has never restored a full application into a replacement account/Region. What practice establishes actual recoverability?

Check the backup plan description and assume all future restores work.
Disable encryption on restored data so it imports faster without checking policy.
Run recurring controlled restore drills and verify data integrity, permissions, dependency order and elapsed recovery time.
Perform only a backup successful-job report and a mock cutover without decrypting a restored database or reconnecting its application clients.

Explanation

Correct Answer

A successful backup job confirms capture, not that usable systems can be rebuilt within target RTO.

Incorrect Answers

Configuration existence is not recovery evidence.

Security can be lost without solving KMS and IAM prerequisites.

A job report or dry-run DNS change does not demonstrate usable, decrypted data, dependent infrastructure, or restoration time. Exercise the complete recovery procedure.

Question 373

A database VM keeps transaction log and data on separate EBS volumes. To recover a coordinated point in time without manually pausing every volume snapshot at different moments, what should be considered?

Start unrelated snapshots hours apart and assume they represent one transaction-consistent state.
Use only EBS volume tags as proof of identical data-block time.
Take a crash-consistent multi-volume snapshot set and assume it automatically flushes every uncommitted in-memory database transaction to storage.
Use EC2 multi-volume snapshots to capture a crash-consistent set, plus application-level quiescence when stronger consistency is required.

Explanation

Correct Answer

Coordinated multi-volume snapshots cover disks of an instance at a crash-consistent point; application semantics may still need additional controls.

Incorrect Answers

Independent capture times can split related database changes.

Tags identify volumes but do not coordinate snapshot state.

Coordinated EBS snapshots provide crash consistency across volumes; an application transaction-consistent recovery point may still require quiescence, database logs or recovery replay.

Question 374

An operator accidentally modified thousands of DynamoDB items this morning. PITR was enabled, and the table should be recovered to just before the change while preserving the current table for investigation. What is the correct action?

Point DynamoDB PITR at the current table and expect all records to roll back in place.
Read only DynamoDB Streams beyond its bounded retention to regenerate every historical item.
Restore the DynamoDB table to a new table at an eligible point-in-time, then validate data and plan cutover.
Delete the existing table first to make PITR available.

Explanation

Correct Answer

DynamoDB PITR creates a new table; it does not rewrite the existing table in place.

Incorrect Answers

PITR restore produces a different target table.

Streams are change records and may not cover the needed history.

Destruction before testing a new restore risks losing forensic evidence.

Question 375

A critical business application is spread across EC2, RDS and Route 53. Architects want an AWS service to evaluate declared recovery targets against resources and resilience recommendations. Which service is appropriate?

Treat Cost Explorer as an RTO simulator for every application dependency.
Run an AWS Backup plan successfully and assume the result calculates every RTO and RPO weakness for the full EC2/RDS/Route 53 application.
Choose an IAM permissions boundary as an automated failover simulator.
Use AWS Resilience Hub to assess a modeled application against RTO/RPO goals and review its recommended improvements.

Explanation

Correct Answer

Resilience Hub evaluates workload recovery posture but is not itself a data replication or backup executor.

Incorrect Answers

Cost reporting does not validate full recovery workflows.

Successful backups are a recovery-data input; they do not model all supported application dependencies or simulate workload recovery against policy. Resilience Hub performs the assessment.

IAM boundaries constrain permissions, not resilience score assessments.

Question 376

A team claims its web system survives AZ failure but has never tested target loss. It wants a controlled experiment that stops eligible EC2 capacity and observes service SLO recovery. Which tool fits?

Disable all alarm guardrails and terminate every production resource simultaneously.
Use AWS Budgets to simulate actual AZ network impairment.
Use AWS Fault Injection Service with approved scope and safety stop conditions, then measure actual recovery.
Run an approved FIS experiment with a stop-condition CloudWatch alarm disabled, relying on a later screenshot to guarantee customer impact remained in bounds.

Explanation

Correct Answer

FIS can inject controlled fault scenarios to challenge system resilience under governed experiments.

Incorrect Answers

Unbounded chaos is unsafe and does not meet controlled experiment requirements.

Cost threshold alerts are not fault injectors.

A fault-injection exercise must include a controlled blast radius and suitable stop conditions with actual telemetry. A post-hoc image is not a safety interlock.

Question 377

A production stack runs in one Region with an idle DR blueprint in another. The DR runbook assumes launching 300 EC2 vCPUs, but the target Region quota is 96. Which prerequisite must be handled?

Assume the primary Region’s quotas automatically apply to all secondary Regions.
Copy the primary-Region CloudFormation template into the DR Region but assume template validation allocates the Region-specific 300-vCPU quota.
Verify and raise secondary-region service quotas and capacity feasibility before claiming the DR design can scale.
Set Route 53 failover to the target Region before verifying it can launch hosts.

Explanation

Correct Answer

Regional quotas and available resource capacity can prevent standby fleet provisioning even when infrastructure templates are correct.

Incorrect Answers

Many AWS service quotas are Region-specific.

Correct templates can fail to launch at quota limits. Service Quotas and representative recovery-scale capacity testing must be checked in the destination Region.

DNS cutover cannot create constrained compute capacity.

Question 378

A load-balanced web fleet intermittently returns 502 errors from one instance that no longer responds to application health requests. How should ALB stop routing to it?

Configure an appropriate target-group application health check and replace or repair the failing instance.
Keep the malfunctioning EC2 target registered but change ALB health-check success codes to include its ongoing 502 errors, treating those failures as healthy.
Disable target health checking and keep sending new traffic to the broken instance.
Rely on Route 53 health of the public ALB as proof every individual target is healthy.

Explanation

Correct Answer

ALB marks unhealthy targets and routes to other available healthy targets, subject to target-group behavior.

Incorrect Answers

Broadening accepted health status to include server errors can hide failed target behavior. Probe a valid application health path and repair or replace unhealthy instances.

That would prolong failures rather than isolate the target.

A healthy ALB hostname does not replace per-target probes.

Question 379

A transactional application needs to survive AZ storage faults while using Aurora. Which database-level characteristic reduces dependence on a single EBS disk/AZ?

Aurora’s distributed cluster storage automatically replicates data across multiple AZs and repairs failed copies.
Store the only cluster volume on the current writer’s ephemeral instance disk.
Disable all automated storage copies and rely on a daily CSV export.
Add an Aurora reader replica in the same AZ as the writer but disable distributed cluster storage redundancy, expecting reader compute alone to protect the volume.

Explanation

Correct Answer

Aurora separates shared resilient cluster storage from compute instances, so storage redundancy spans AZs.

Incorrect Answers

Aurora cluster volume is not only the writer’s local disk.

That sacrifices low-RPO availability.

Reader compute is a failover candidate, but Aurora durable cluster storage is a distinct service layer replicated across Availability Zones, not an ephemeral copy on one host.

Question 380

An RDS system uses automated backups and Multi-AZ failover, with retention required for a compliance recovery window. What should its operational plan verify?

Assume Multi-AZ standby is a permanent immutable backup of every deleted row.
Disable automated backups immediately after the first healthy failover.
Promote a Multi-AZ standby during an outage and assume that the standby retains every historical row deleted during the past 30 days.
Confirm automated-backup retention and restorable points independently of failover, then test restoring to a separate instance.

Explanation

Correct Answer

Failover continuity and historical recovery are different capabilities; a successful standby takeover does not prove older points remain usable.

Incorrect Answers

Standby replication propagates writes and deletions, so historical recovery needs backups.

That removes required historical recovery points.

Standby replication includes recent writes and deletes. Recovering a past version of data requires backups and supported point-in-time restore history.

Pass Your Amazon AWS Certified Solutions Architect - Associate SAA-C03 Exam Easy!

Question 370

Explanation

Correct Answer

Incorrect Answers

Question 371

Explanation

Correct Answer

Incorrect Answers

Question 372

Explanation

Correct Answer

Incorrect Answers

Question 373

Explanation

Correct Answer

Incorrect Answers

Question 374

Explanation

Correct Answer

Incorrect Answers

Question 375

Explanation

Correct Answer

Incorrect Answers

Question 376

Explanation

Correct Answer

Incorrect Answers

Question 377

Explanation

Correct Answer

Incorrect Answers

Question 378

Explanation

Correct Answer

Incorrect Answers

Question 379

Explanation

Correct Answer

Incorrect Answers

Question 380

Explanation

Correct Answer

Incorrect Answers

AWS Certified Solutions Architect - Associate SAA-C03 Practice Tests