Question 670
A migrated claims processor executes multiple sequential external API steps, some of which may take minutes and need retries with compensation on failure. Operators currently coordinate steps by custom timeouts embedded in Lambda handlers. Which modernization best externalizes workflow state?
- Use separate EventBridge scheduled rules to invoke each external API step on a static timer, without durable step success tracking or compensation branches.
- Chain Lambda calls and keep long-running workflow state only in a single API process, assuming compensation and retries survive process shutdown automatically.
- Use AWS Step Functions with a suitable state machine, explicit retries, timeouts and compensating actions for the long-running process.
- Run all steps in a single Lambda invocation using local variables to preserve the workflow, assuming executions can exceed the service timeout without interruption.
Explanation
Correct Answer
C:Step Functions represents durable workflow progress and retry/branch behavior outside individual Lambda invocations, enabling auditable orchestration.
Incorrect Answers
A:Schedules are time triggers rather than a state machine that records step completion, branches, retries and compensating actions during long-lived claims workflows.
B:Lambda and process-local state are not a durable end-to-end workflow controller; Step Functions owns durable state transitions and retry/compensation logic.
D:Lambda invocation duration and memory state are bounded; Step Functions externalizes durable progress and controls retries, wait states and compensation.
Question 671
An ecommerce system wants each new order event delivered independently to fraud, shipment and analytics consumers. Consumers should evolve at different rates, and the order service must not know their endpoints. Which event architecture fits?
- Publish an order event to one SQS queue and let all three consumer teams compete to receive messages from that same queue.
- Configure one FIFO queue with three consumer worker groups competing for the same order message, assuming each worker group receives its own copy.
- Configure EventBridge rules but deliver all matching orders into a shared single-consumer queue used by one combined application.
- Publish order events to an EventBridge event bus and route matching events to independently configured service targets.
Explanation
Correct Answer
D:EventBridge rules decouple producers from multiple consumers and allow separate subscriptions and delivery behavior for matching events.
Incorrect Answers
A:SQS competing consumers divide messages rather than deliver each event independently to fraud, shipping and analytics. A bus with separate rules/targets supports fanout.
B:A queue delivers each message to one consuming worker, not a copy to every separate domain; independent EventBridge targets or fanout queues are needed.
C:A single shared queue still couples independently evolving consumers, rather than providing distinct delivery subscriptions and retry policies for each domain.
Question 672
A modernization team needs messages from a financial transaction producer preserved until independent workers process them. Worker outages can last hours and messages must support retries and a dead-letter queue. Which service semantics best fit?
- Use Amazon SNS with no durable subscribed SQS queues and treat topic publication alone as a worker message queue with visibility timeout and acknowledgement.
- Use an EventBridge event bus with short-lived default event retention as a replacement for long-running independent worker queues and explicit receive acknowledgments.
- Publish a transaction event onto EventBridge and expect a receiving worker to delete it with SQS DeleteMessage after successful processing even though no SQS queue exists.
- Use Amazon SQS queues with suitable retention, visibility timeouts, redrive policies and idempotent consumers.
Explanation
Correct Answer
D:SQS durably buffers work for pull consumers and supports retry, DLQ and worker decoupling during outages.
Incorrect Answers
A:SNS offers event fanout but not the queue receive/visibility/redrive semantics required for consumers offline for hours unless durable queues are subscribed.
B:An EventBridge event bus routes events; it is not a durable pull work queue with SQS visibility and redrive to meet the specified worker semantics.
C:Event buses are not pull queues and do not expose SQS receive/visibility/delete-message semantics; use an actual durable SQS work queue.
Question 673
A legacy monolith runs monthly PDF jobs by polling an RDS table every second. The new architecture should run code only when files arrive in a designated S3 prefix, with no dedicated server fleet. Which modernization is most direct?
- Trigger an appropriately configured Lambda processor from S3 object-created events, with idempotency and failure handling.
- Subscribe Lambda to all S3 object events, including its own generated output in the same bucket and prefix, without a filter or idempotency guard.
- Continue second-by-second RDS polling but invoke Lambda from a scheduled EventBridge rule, keeping the polling frequency and query cost.
- Trigger Lambda from every object-created notification in a bucket where its output uses the same prefix, assuming no duplicate or recursive invocations can occur.
Explanation
Correct Answer
A:S3 events can drive serverless processing on arrival, removing wasteful database polling and always-on compute for intermittent file tasks.
Incorrect Answers
B:Unfiltered self-triggering events can cause processing loops. The event scope and destination must not recursively invoke the function on its own output.
C:Moving polling into scheduled functions does not provide event-driven execution on the specified S3 object arrival; S3 notifications directly trigger work.
D:An unfiltered notification can trigger recursively when output is written; filter the input prefix and use idempotent failure handling.
Question 674
A customer profile API currently requires a large EC2 server fleet to handle sporadic HTTP requests. The core business logic is stateless, each request completes quickly, and the team wants managed request routing, authorization and no idle web hosts. Which service combination is suitable?
- Move the web application behind API Gateway but forward every request to the same permanently running EC2 fleet as the Lambda-free integration target.
- Use Amazon API Gateway with Lambda integrations and appropriate authorization, throttling and observability controls.
- Place a permanently running ECS service behind API Gateway and assume it eliminates idle compute cost for sporadic traffic without any task scale-down.
- Run Lambda behind an ALB but omit necessary authorization and throttling while assuming the ALB alone satisfies every managed API requirement.
Explanation
Correct Answer
B:API Gateway handles HTTP API entry while Lambda supplies event-driven stateless execution without an always-running EC2 fleet.
Incorrect Answers
A:API Gateway alone does not remove idle web hosts or EC2 fleet maintenance; integrating stateless logic with Lambda meets the serverless requirement.
C:ECS with nonzero always-on capacity can serve HTTP, but does not meet the stated no-idle-host execution preference; Lambda with API Gateway does.
D:ALB-to-Lambda can execute serverlessly but does not automatically supply the specified managed API authorization and usage control configuration.
Question 675
A retail application uses DynamoDB Streams to track order records. The DynamoDB stream view includes the new image of modified items; downstream fulfillment wants only events whose NEW order status is SHIPPED, not every update. It must deliver qualifying stream records to an SQS queue, optionally enrich them with another service, and avoid a custom Lambda stream poller. Duplicate deliveries can be handled idempotently. Which integration fits?
- Attach the DynamoDB stream to a CloudFront cache behavior and assume edge cache settings implement state-image filtering.
- Use an EventBridge Pipe with the DynamoDB stream as source, a filter on NewImage status, optional enrichment, and SQS as an authorized target.
- Use a scheduled full DynamoDB table scan and assume the resulting snapshot preserves each distinct order update event for downstream consumers.
- Create an EventBridge event-bus rule with no pipe or other stream source and assume the bus automatically polls all DynamoDB stream shards.
Explanation
Correct Answer
B:EventBridge Pipes can poll DynamoDB Streams, apply supported event filtering on the new image, optionally enrich matching records, and deliver to SQS using a scoped execution role; downstream delivery is at least once.
Incorrect Answers
A:CloudFront request cache behaviors do not consume DynamoDB stream images or forward their selected changes as messages.
C:A table scan sees current item state rather than the ordered stream of change records and can miss intermediate status updates.
D:An event-bus rule matches events already available to the bus; it does not automatically attach to an unrelated DynamoDB stream without a supported source integration.
Question 676
A company modernizes an event bus shared by independently deployed microservices. Producers add optional fields to order events every few weeks, and consumers want a discoverable contract and typed code bindings without tightly coupling their release schedules. What AWS-native capability addresses this particular interoperability need?
- Use EventBridge Schema Registry and compatible schema discovery or versioning practices to publish event shapes and generate bindings for supported consumer tooling.
- Rely only on EventBridge rules matching detail-type and assume those rules independently publish typed schemas and code bindings for all consumers.
- Make every event publisher call every consumer synchronously so all services must deploy their schemas at the same moment.
- Copy example event payloads into a shared S3 object without managing a discoverable schema contract or compatible version changes.
Explanation
Correct Answer
A:EventBridge Schemas can surface event structures and support schema-based code generation, while producers and consumers still need an explicit backwards-compatibility discipline.
Incorrect Answers
B:Routing rules deliver selected events but do not provide schema discovery, registry versions or generated consumer bindings.
C:Synchronous point-to-point calls increase release coupling and do not solve schema compatibility when fields change.
D:Sample objects are not the schema catalog and code binding workflow requested by independently evolving consumers.
Question 677
A fulfillment platform has already decoupled its API from a durable SQS queue. Its ECS worker service uses a fixed desired task count, so backlog grows sharply during promotions even though each task has predictable processing throughput. CPU stays moderate. What scaling signal is most closely tied to the actual queueing delay?
- Scale worker ECS tasks only on CPU target tracking even though measured CPU remains flat while backlog age rises.
- Configure ECS Service Auto Scaling from a validated backlog-per-running-task metric or appropriate workload demand metric, with capacity limits and target processing latency.
- Increase SQS message retention and keep the worker desired count fixed, expecting retention configuration to raise ECS capacity.
- Turn off the SQS DLQ so that poison messages can continuously retry and inflate the queue-depth metric.
Explanation
Correct Answer
B:Backlog per task relates pending work to available processing concurrency; a properly scaled ECS service can increase consumer capacity to meet latency objectives, subject to resource and downstream limits.
Incorrect Answers
A:CPU is not correlated with this queue-bound demand; backlog per running task or processing-age metrics better reflect delay.
C:Longer retention helps avoid expiry but does not create additional workers or reduce current queue wait time.
D:Disabling failure isolation increases retries and does not create reliable demand-based capacity control.
Question 678
A migration utility must run a short data-quality function against each of 900,000 objects listed in S3. The input exceeds ordinary state payload limits, item-level failures must be visible, and the team needs a configurable concurrency ceiling. Which orchestration mechanism best fits the scale?
- Use an Inline Map with all 900,000 objects passed through the parent state and assume it bypasses state payload and execution-history limits.
- Use a basic Step Functions Inline Map state without checking its payload, event-history and concurrency bounds.
- Use a Step Functions Distributed Map reading a supported S3 dataset or item source, controlling child-workflow concurrency and failure thresholds.
- Write all 900,000 S3 keys into one state input without an item source and assume Step Functions removes its payload limits.
Explanation
Correct Answer
C:Distributed Map is designed for high-concurrency processing from large S3 datasets with child executions, failure thresholds, and a managed Map Run for monitoring.
Incorrect Answers
A:Inline Map has payload, history and concurrency limits that this scale explicitly exceeds.
B:Inline Map has smaller concurrency, input, and history limits than the specified large-scale dataset requires.
D:The workflow requires a supported S3 item reader and distributed child executions rather than a single oversized inline payload.
Question 679
A migrated document analyzer runs in an OCI container and requires a 25-minute CPU-intensive process with 2 vCPU and 8 GiB memory. Runs are sporadic, do not require GPUs, and should not tie up an incoming HTTP connection. The team wants no worker EC2 fleet. What compute target is most appropriate?
- Use a single AWS Lambda invocation to execute the 25-minute task and assume the function duration limit can be extended to cover it.
- Use a provisioned AWS Batch EC2 compute environment with always-on worker hosts, even though sporadic jobs can use supported on-demand Fargate tasks without hosts.
- Accept the request asynchronously and launch an ECS task on AWS Fargate with the required supported CPU and memory configuration, recording progress and completion separately.
- Use API Gateway request integration timeouts as if extending them runs the 25-minute container without any compute service.
Explanation
Correct Answer
C:Fargate supports container workloads without customer-managed EC2 hosts; asynchronous execution avoids keeping clients connected for a long-running task.
Incorrect Answers
A:Lambda has a hard per-invocation runtime ceiling below 25 minutes; ECS on Fargate can execute the eligible container asynchronously.
B:Batch EC2 could run the work but introduces EC2 worker management that the request explicitly seeks to avoid.
D:API integration timeouts only govern the request path and cannot execute a native container workload with no backend.
Question 680
A large monolith must gradually move individual functions to managed services while retaining the existing public domain and supporting both old and new paths during the transition. What migration pattern reduces big-bang risk?
- Replace every public API route simultaneously with new services and remove access to the legacy monolith before validating the replacement paths.
- Disable all monitoring during migration so traffic-cutover regressions cannot stop service extraction.
- Apply a strangler pattern using API routing to shift selected capabilities behind stable interfaces incrementally.
- Keep all external traffic routed permanently to the original monolith, while developing new services that never receive production paths.
Explanation
Correct Answer
C:Incremental routing and service extraction allow tested functionality to move without cutting over the entire monolith at once.
Incorrect Answers
A:A big-bang switch defeats incremental strangler migration and increases rollback risk.
B:Removing telemetry conceals defects and makes staged rollout and rollback less reliable.
D:Without controlled routing transfer, the planned incremental extraction cannot retire old responsibilities or validate the new services.
AWS Certified Solutions Architect - Professional SAP-C02 Practice Tests