Amazon AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 35

Question 340

After changing a synchronous image API to an SQS-backed pipeline, which observation best confirms that HTTP requests can succeed while workers are temporarily stopped?

The uploader returns success only after the image thumbnail finishes rendering.
A test upload is accepted and placed on the target queue, then processes successfully after workers resume without resubmitting the HTTP request.
The S3 bucket has the intended name but no message or output appears.
Worker CPU falls when workers are stopped and the upload endpoint also fails immediately.

Explanation

Correct Answer

The experiment directly demonstrates producer/consumer decoupling and durable pending work.

Incorrect Answers

That remains synchronous dependence rather than proof of queue buffering.

Resource naming alone does not show accepted work survives a consumer outage.

That indicates the failure remains coupled to request acceptance.

Question 341

A team deploys SQS FIFO groups by customer. Which test demonstrates both intended guarantees rather than merely confirming the queue exists?

Verify only that total queue size drops to zero after a single request.
Send ten messages all from one group and conclude cross-group concurrency works.
Inspect the S3 bucket replication status instead of the received sequence.
Interleave two customers’ numbered updates, verify each customer’s order is preserved while both customer groups make progress concurrently.

Explanation

Correct Answer

This measures per-group ordering and independent group throughput under realistic interleaving.

Incorrect Answers

An empty queue says little about ordering between successive events.

One group cannot test independence across groups.

Storage replication does not validate FIFO consumer ordering.

Question 342

A worker extends SQS visibility for 10-minute jobs. Which test directly shows the extension prevents premature redelivery, without claiming exactly-once processing?

Assume configured visibility means the same SQS message can never be delivered twice.
Test only a job that finishes before the original timeout.
Check only the primary queue approximate message count when the worker reports a visibility extension, without testing an actual job exceeding the original timeout.
Run a job past the original timeout, extend visibility before expiry, and inspect receive counts and concurrent worker logs while it remains in flight.

Explanation

Correct Answer

The test checks effective invisibility under real duration and still acknowledges rare at-least-once redelivery possibilities.

Incorrect Answers

Delivery guarantees remain at least once and duplicate-safe processing is required.

That does not exercise the extension path.

Queue count alone cannot verify in-flight invisibility under extended work. Exercise a job longer than the first timeout and verify receive/worker evidence while it runs.

Question 343

A poison JSON message has failed repeatedly in a staging queue after a new redrive policy. What evidence proves the intended recovery boundary works?

Observe one handler exception and assume the message was automatically isolated.
Only check that the main queue name exists in AWS Console.
Show the message reaches the configured DLQ at the maxReceiveCount threshold, remains inspectable and can be deliberately redriven after a fix.
Increase the consumer batch size and treat fewer polling cycles as proof invalid JSON reaches the DLQ after maxReceiveCount.

Explanation

Correct Answer

This validates both quarantine and controlled recovery rather than only a configuration flag.

Incorrect Answers

A failure alone does not prove DLQ association or threshold behavior.

An existing queue says nothing about redrive.

Batch sizing affects consumption throughput, not whether a poison record is routed to the configured DLQ and can be redriven. Test receive threshold and destination.

Question 344

Following an SNS fanout deployment, how can the team prove that three downstream departments get independent copies and retry state?

Check that exactly one shared SQS queue received the message.
Validate only the SNS topic name, without observing the subscribed queues.
Connect each downstream department as a competing consumer of the same SQS queue and expect one SNS publish to create independently retryable copies for each department.
Publish one test event, verify its identifier arrives separately in all three subscribed queues and each team can process it despite one consumer being paused.

Explanation

Correct Answer

Per-queue receipt and independent backlog demonstrate true fanout and isolation.

Incorrect Answers

One common work queue does not prove every department got a copy.

Topic existence is not delivery evidence.

A shared work queue delivers individual messages to competing receivers; distinct subscriptions to separate queues are required for each department to own its backlog.

Question 345

A team adds EventBridge pattern filtering for SaaS order events. Which test confirms the rule does not forward unrelated events?

Call the target Lambda directly and treat its success as proof the rule matches.
Check only a CloudWatch CPU graph for the target’s EC2 host.
Publish no events and conclude the rule filters all traffic correctly because there are zero invocations.
Publish matching and nonmatching events to the intended bus and inspect matched-rule invocations and target receipts for each.

Explanation

Correct Answer

Positive and negative controls prove event-pattern accuracy, not just that the target can run.

Incorrect Answers

Direct invocation bypasses EventBridge matching.

CPU cannot prove event classification or rule invocation.

An idle result cannot distinguish correct filtering from a broken rule.

Question 346

Operations replaces a VM crontab with EventBridge Scheduler for a monthly task. What direct verification establishes invocation is no longer dependent on the VM staying online?

Confirm the VM CPU falls when crontab is disabled, without examining task execution.
Keep the old VM crontab active and check that the target Lambda exists, assuming the Lambda name proves EventBridge Scheduler invoked it independently.
Disable the old VM schedule, trigger an approved test schedule and inspect Scheduler delivery/target execution and retry receipts.
Start a new local crontab on a second VM and label it EventBridge.

Explanation

Correct Answer

The new managed schedule should deliver to the intended target independently of the previous host.

Incorrect Answers

CPU idle is not proof the managed schedule works.

A function existing does not establish managed scheduling or isolate the host timer. Disable the old crontab and test Scheduler target invocation/receipts.

That still relies on a self-managed host timer.

Question 347

A Step Functions payment state machine was added to replace manual orchestration. Which test validates the intended ordering and failure-compensation policy?

Inject failures after validation and after reservation, inspect execution history, and verify that no invalid commit occurs and the defined compensation runs only when required.
Only run a case in which every task succeeds and assume all Catch paths work.
Count how many Lambda functions exist in the account.
Check only that a Step Functions definition was published, without injecting failures or inspecting the actual Catch, Retry and compensation branch behavior.

Explanation

Correct Answer

A controlled negative-path test proves order and recovery, not merely successful happy-path status.

Incorrect Answers

Happy-path success does not exercise conditional error transitions.

Function inventory does not validate state ordering or outcomes.

Definition existence or a happy path is not evidence that after-partial-commit failures execute the intended compensating operations. Run targeted fault-injection paths and inspect history.

Question 348

A design chooses Step Functions Standard for a multi-day approval. What evidence directly distinguishes it from an unsuitable Express configuration in the deployed state?

Confirm an Express workflow completes a 10-second trivial example.
Inspect execution type, active long-lived state history and approved wait/restart behavior using a controlled approval delay.
Assume a Lambda timeout value proves the state machine type is Standard.
Configure a Lambda timeout of 15 minutes and assume that setting permits Express state-machine executions to wait for multi-day approval.

Explanation

Correct Answer

Workflow type and actual delayed execution verify the duration and audit requirements beyond naming a state machine.

Incorrect Answers

A short successful run cannot establish multi-day durability.

A function timeout does not identify workflow execution semantics.

The task Lambda timeout cannot change a Step Functions Express execution-duration contract. Choose Standard for the long-lived state and verify a delayed execution.

Question 349

An internet-facing web application needs to remain available after any one Availability Zone fails. Where should its ALB targets and autoscaling instances be placed?

Across at least two independent AZ subnets, with ALB health checking and appropriately sized target capacity.
Only one subnet/AZ with many instances.
Configure an ALB with cross-zone routing but register healthy application targets only in AZ-a and assume cross-zone routing creates servers in AZ-b.
A single database standby in another AZ instead of web target redundancy.

Explanation

Correct Answer

Regional ALB routing can avoid unhealthy AZ targets when healthy capacity exists in other enabled zones.

Incorrect Answers

Instance redundancy inside one AZ does not protect against zone failure.

Cross-zone forwarding cannot manufacture compute in a lost AZ. The target fleet itself must run in multiple available AZs with sufficient remaining capacity.

Database HA does not supply replacement web targets.

Question 350

A relational database needs automatic availability after a primary instance/AZ outage, but analytic read scaling is not required. Which RDS deployment meets that requirement?

An ordinary read replica with no controlled promotion or failover procedure.
Add an RDS snapshot policy with hourly automated backups and assume those restore points replace automatic failover during an AZ outage.
A larger single-AZ database instance without standby.
An RDS Multi-AZ DB instance deployment with a synchronous standby and automated failover.

Explanation

Correct Answer

A Multi-AZ DB instance standby protects availability; it is not a read-only application endpoint.

Incorrect Answers

Asynchronous replicas are not identical automatic Multi-AZ HA standbys.

Backups support historical restore but require provisioning and a controlled cutover, rather than automatic standby promotion in a traditional Multi-AZ DB instance.

More resources do not remove the AZ single point of failure.

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

Question 340

Explanation

Correct Answer

Incorrect Answers

Question 341

Explanation

Correct Answer

Incorrect Answers

Question 342

Explanation

Correct Answer

Incorrect Answers

Question 343

Explanation

Correct Answer

Incorrect Answers

Question 344

Explanation

Correct Answer

Incorrect Answers

Question 345

Explanation

Correct Answer

Incorrect Answers

Question 346

Explanation

Correct Answer

Incorrect Answers

Question 347

Explanation

Correct Answer

Incorrect Answers

Question 348

Explanation

Correct Answer

Incorrect Answers

Question 349

Explanation

Correct Answer

Incorrect Answers

Question 350

Explanation

Correct Answer

Incorrect Answers

AWS Certified Solutions Architect - Associate SAA-C03 Practice Tests