AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 48
Question 470
A document-validation workflow has completed three external service calls when a fourth fails. Operators need to resume safely without sending duplicate invoices from earlier stages. What must the state design preserve?
- Track durable progress and idempotent step boundaries so retries or resumes do not repeat already committed side effects.
- Configure unbounded Step Functions retries for every failed task, but omit business-operation idempotency keys for the invoices that previous tasks already issued.
- Make every external service request unconditional because Step Functions guarantees exactly-once effects in outside systems.
- Restart the workflow from the beginning after any failure and rely on the execution name alone to prevent previously committed invoice API calls from running again.
Explanation
Correct Answer
A:State-machine transitions and application idempotency are needed because re-execution across non-atomic services can otherwise resend earlier invoices.
Incorrect Answers
B:Step Functions retry behavior does not provide transaction atomicity for external billing systems. Without durable committed-step identifiers, retry/redrive can repeat a successful side effect.
C:Workflow execution does not atomically coordinate every independent provider's business write.
D:Execution naming may help identify workflow attempts, but does not make an external invoice endpoint idempotent. Durable per-step progress and operation IDs prevent duplicate effects on replay.
Question 471
A pricing service makes a synchronous request to a tax API because the final quoted price must be shown within 300 milliseconds. The tax API normally replies in 40 milliseconds. What interaction model is suitable if failure handling is designed?
- Store tax API request bodies in a CloudWatch metric and assume the metric returns computed tax synchronously.
- Enqueue tax calculations for tomorrow and return a final supposedly correct price without knowing applicable tax.
- Use a bounded synchronous request-response call with strict timeout and fallback behavior.
- Use a multi-hour Step Functions human approval for every quote even though no approval is required.
Explanation
Correct Answer
C:When the downstream result is genuinely needed to form the immediate response and the latency budget supports it, a synchronous dependency can be appropriate.
Incorrect Answers
A:Metrics store observations and are not a substitute for the tax calculation API response.
B:A delayed calculation cannot satisfy a price that must include current tax before responding.
D:Unneeded long orchestration adds latency rather than meeting the real-time response condition.
Question 472
An order service publishes 'order confirmed' notifications for recommendations and shipment analytics. Neither downstream consumer needs to finish before checkout completes. What integration reduces coupling?
- Wait synchronously for every analytics and recommendation service before acknowledging each payment.
- Run all downstream consumer business logic inside the user's browser after returning payment success.
- Use an ALB listener rule to guarantee that every independent consumer receives the same event exactly once.
- Publish an asynchronous event through EventBridge or SNS with independent downstream subscriptions.
Explanation
Correct Answer
D:Nonblocking event delivery lets checkout commit without depending on the availability or latency of unrelated analytics consumers.
Incorrect Answers
A:Unnecessary synchronous waits make checkout reliability depend on slower optional processing tiers.
B:Browser-side optional work cannot be relied on as a durable authenticated backend processing pipeline.
C:Load balancing distributes requests among targets rather than implementing durable multi-consumer event fan-out.
Question 473
A sensor collector ingests 25,000 readings per second while a validation service processes only 4,000 per second during maintenance. The collector must avoid backpressure timeouts. What architecture absorbs the mismatch?
- Delete excess readings during maintenance because successful receipt implies they are already processed.
- Use Route 53 weighted routing as the sole durable event buffer for readings.
- Make the producer call each validator synchronously and keep all HTTP connections open for hours.
- Buffer readings in a suitable durable queue or stream, then scale consumers against backlog and processing capacity.
Explanation
Correct Answer
D:Decoupling the arrival rate from processing lets the ingestion endpoint return promptly while workers catch up as capacity permits.
Incorrect Answers
A:Dropping unprocessed events violates data preservation and confuses acceptance with completed work.
B:DNS traffic selection does not store sensor payloads for later consumption or replay.
C:Long synchronous waits couple availability to constrained consumers and cannot reliably absorb sustained burst traffic.
Question 474
A shopping API submits a payment job to SQS and immediately tells the user 'payment completed' even though the worker has not processed the message. Customers later report missing charges. What should the user-facing contract distinguish?
- Separate 'accepted for processing' from 'completed' and track durable outcome using a correlation or job identifier.
- Change SQS message retention to fourteen days and continue returning a false completed status immediately.
- Treat every successful SendMessage response as evidence the payment provider irreversibly settled the charge.
- Remove all correlation IDs because messages in SQS are globally ordered and never retried.
Explanation
Correct Answer
A:A queue acknowledgement confirms submission, not the downstream business transaction. A job status model and correlation ID make eventual success or failure observable.
Incorrect Answers
B:Retention supports recovery time but does not turn pending asynchronous processing into completed payment work.
C:Message acceptance occurs before the consumer's independent payment operation, which can fail or be delayed.
D:Queue order is not a substitute for tracking which request produced a later success or failure result.
Question 475
A team promises its database-heavy new API will automatically handle 100,000 transactions per second because the EC2 application servers use Auto Scaling. What scaling prerequisite is missing?
- Ignore all account quotas because adding an Auto Scaling policy removes managed-service rate limits permanently.
- Review relevant AWS service quotas, database limits and downstream throughput before committing to the target rate.
- Change the application DNS name to a shorter string to raise RDS write transaction capacity.
- Set larger EC2 root disks and assume database connection and throughput restrictions disappear.
Explanation
Correct Answer
B:Auto Scaling of compute does not bypass API limits, data-service partition constraints or regional resource quotas on dependencies.
Incorrect Answers
A:Regional and per-service quotas remain enforceable regardless of the number of clients requesting resources.
C:DNS naming has no causal relationship to relational database write throughput or API quotas.
D:Instance root disk size does not increase a separate database service's connection or write limits.
Question 476
A regional service quota allows far fewer new ENIs than planned for a major Fargate expansion next week. The architecture review discovers the cap only after deployment has been scheduled. What should occur before launch?
- Disable CloudWatch quota metrics so the expansion request is no longer blocked by monitoring alarms.
- Increase API Gateway request timeout to make the Fargate account limit disappear.
- Request an eligible quota increase or redesign the resource footprint and validate realistic capacity in advance.
- Proceed without change because ECS Service Auto Scaling automatically cancels every regional quota when demand rises.
Explanation
Correct Answer
C:An identified quota can prevent provisioning even if the application design is otherwise elastic. Preflight and increase requests reduce surprise scale failures.
Incorrect Answers
A:Metrics describe usage; disabling them cannot change a quota or resource-provisioning permission.
B:An HTTP timeout setting is unrelated to the maximum number of provisioned network resources.
D:Service autoscaling cannot create resources past hard account or regional provider quotas.
Question 477
Hundreds of Lambda workers call an AWS API at once after an SNS fan-out event. Requests are throttled, and the current code retries immediately in lockstep, causing another spike. How should retry behavior change?
- Use capped exponential backoff with jitter for throttled requests and respect service-specific retry guidance.
- Treat HTTP 429 as irreversible success and silently discard the requested state change.
- Retry each throttled request in a tight loop with no delay until the remote API accepts it.
- Double the number of workers on each throttling response without changing request rate logic.
Explanation
Correct Answer
A:Exponential delays reduce retry pressure while random jitter spreads workers over time instead of causing synchronized retry storms.
Incorrect Answers
B:A throttle response indicates the action was not accepted as requested and needs controlled retry or failure handling.
C:Immediate repeated calls maintain overload and increase throttling and wasted compute.
D:More concurrent callers can make the throttled dependency even less able to recover.
Question 478
A region runs both customer authorization Lambda functions and optional batch clean-up tasks. The batch jobs occasionally consume all available concurrency and impact logins. What isolation approach can be considered?
- Give background jobs unlimited concurrency and rely on average account CPU to protect login requests.
- Disable all access logs from the login service because smaller logs increase Lambda account concurrency quota automatically.
- Apply reserved concurrency or other supported capacity limits to critical and background functions based on priorities.
- Move payment authentication code into an S3 lifecycle rule so it runs outside application compute limits.
Explanation
Correct Answer
C:Resource isolation can protect a latency-sensitive path from noisy neighbors in the same concurrency pool, while limiting nonessential parallel work.
Incorrect Answers
A:Shared execution concurrency can still be exhausted even when CPU utilization is not the controlling limit.
B:Logging volume and function concurrency quotas are independent resource considerations.
D:Object lifecycle transitions cannot execute arbitrary sensitive user authorization logic.
Question 479
A web fleet scales out and each EC2 instance maintains its own login-session cache. Users lose session data when the ALB changes targets. The business wants low-latency shared access. Which AWS caching model is suitable?
- Increase ALB HTTP header size to serialize all server-side login state inside the network listener automatically.
- Use a suitably replicated Amazon ElastiCache cluster for shared sessions, with TTL and failure handling.
- Use EC2 instance store as one globally shared multi-AZ session volume without deploying a shared file protocol.
- Continue using a separate process-local cache per EC2 host and disable all shared writes.
Explanation
Correct Answer
B:A managed shared in-memory cache allows multiple stateless web hosts to read the same session state independent of load-balancer target selection.
Incorrect Answers
A:Header limits do not transform an instance-local session cache into a shared durable service.
C:Instance-local storage does not transparently expose the same data to independent fleet members across AZs.
D:Local-only entries are not available on other targets after replacement or traffic redistribution.
Question 480
A securities dashboard caches current stock availability in Redis for ten hours, even though availability changes every minute. Traders see stale values after stock is exhausted. What cache policy needs correction?
- Use an additional Redis replica without changing TTL or write invalidation, expecting replicas to refresh old values automatically.
- Increase the cache TTL to thirty days so fewer backend requests occur even when stale values cause incorrect trades.
- Disable all cache invalidation events because consistent stale data is safer than frequent refreshes.
- Set suitable cache TTLs and invalidation or update-on-write strategies matched to data freshness requirements.
Explanation
Correct Answer
D:Mutable financial or inventory values require bounded staleness. Caching without timely invalidation can make the application serve incorrect operational decisions.
Incorrect Answers
A:Replication copies the cached state; it does not decide when outdated business information must be expired.
B:Longer retention magnifies data staleness rather than matching the business's minute-level change rate.
C:Avoiding invalidation preserves incorrect values beyond their validity and can violate the specified freshness requirement.
