AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 50
Question 490
An ECS application runs the same image in development, testing and production, but each environment has different database endpoints and API credentials. A developer rebuilds the entire image with values hardcoded for every environment. Which configuration pattern improves control?
- Supply environment-specific settings from managed configuration and secret stores at task startup or runtime.
- Require engineers to SSH into each running Fargate task and edit configuration files manually.
- Use S3 object versioning as the sole runtime secret store without workload IAM permissions or access controls.
- Embed every environment's production password in the public image so any deployment can connect everywhere.
Explanation
Correct Answer
A:Separating deployment settings from image contents allows one validated artifact to be deployed with controlled, environment-specific parameters and credentials.
Incorrect Answers
B:Fargate tasks are ephemeral and managed; manual host editing is not a reproducible configuration deployment.
C:Object versioning is not a replacement for scoped secret retrieval and credential lifecycle management.
D:A shared image containing all secrets creates unnecessary credential exposure and violates environment isolation.
Question 491
A container team patches application code directly inside long-running ECS task containers. When tasks are restarted, modifications disappear and instances run inconsistent versions. What deployment approach is more reliable?
- Use ALB sticky sessions as the mechanism that copies code changes from one container to all others.
- Keep editing each running container interactively because ECS automatically commits changes to the source image.
- Build immutable versioned container images and deploy new task definitions/services referencing the intended image version.
- Disable ECS health replacement so each mutated container survives indefinitely regardless of failures.
Explanation
Correct Answer
C:Container images and task-definition revisions make rollouts reproducible; modifying live task files does not survive orchestrator replacement.
Incorrect Answers
A:Session affinity changes request routing and cannot distribute modified executable images among tasks.
B:Runtime edits are not normally persisted back to an image registry or reused by new tasks.
D:Suppressing replacement undermines recovery and does not create a versioned reproducible artifact.
Question 492
During a rolling ECS service update, old tasks are stopped while they still handle long HTTPS responses, causing truncated client downloads. Which orchestrator/load-balancer behavior should be tuned?
- Immediately terminate all old tasks with no grace period as soon as the new image is registered.
- Increase S3 lifecycle retention because object expiration controls HTTP response draining in ECS.
- Disable ALB target health checks permanently so terminated ECS containers keep receiving requests.
- Allow target deregistration draining and suitable container stop grace periods for in-flight requests.
Explanation
Correct Answer
D:Connection draining and graceful termination let old tasks finish eligible active work before the orchestrator removes their capacity.
Incorrect Answers
A:Abrupt stop interrupts open requests and reproduces the observed user-facing failures.
B:Storage retention cannot delay container termination or preserve a live network connection.
C:Routing to dead targets causes additional failures and is not a graceful rollout mechanism.
Question 493
A stateful container writes customer documents only to its task's writable filesystem. The ECS service replaces the container during deployment and the documents disappear. What persistence change is necessary?
- Disable ECS service autoscaling and assume that prevents infrastructure failure and data loss forever.
- Use a larger container image tag as the durable archive for runtime-created customer documents.
- Increase the task ephemeral-storage allocation and assume that guarantees data survives every task replacement.
- Store durable customer documents in an external persistent service such as S3, EFS or a suitable database.
Explanation
Correct Answer
D:Task-local files should be treated as ephemeral unless explicitly backed by durable external storage with the required sharing semantics.
Incorrect Answers
A:Eliminating planned scaling cannot guarantee an individual host or task survives replacement or failure.
B:An image reference identifies deployment software, not mutable documents generated during an execution.
C:Additional scratch capacity does not change whether a new task reuses the old task's writable filesystem.
Question 494
A public API serves multiple mobile app generations. A new incompatible JSON schema is deployed without versioning, and older clients fail immediately. What API contract technique helps maintain backward compatibility?
- Expose versioned endpoints or contracts with a controlled migration/deprecation process for clients.
- Increase API Gateway throttling to zero; blocked old clients are now considered compatible.
- Change the existing schema in place and force every mobile device to upgrade simultaneously before the next request.
- Shorten the API's DNS record TTL so old mobile parsers can decode the incompatible new JSON fields.
Explanation
Correct Answer
A:Versioning gives independently deployed clients a stable interface while new schemas evolve and compatibility requirements are handled deliberately.
Incorrect Answers
B:Rejecting requests avoids workload but does not provide an usable backward-compatible API contract.
C:Independent client release cycles make simultaneous forced upgrades impractical and risk broad outages.
D:DNS cache timing does not transform incompatible request and response schemas used in client code.
Question 495
A travel platform replaces a monolithic booking backend with several new microservices, but customers still call the old stable API contract. The company cannot force every client integration to change this month. What adaptation layer can help?
- Require all customers to send raw internal DynamoDB SDK requests instead of the published booking API.
- Translate the stable external request/response contract to internal service operations in an API gateway or dedicated adapter layer.
- Remove the existing API hostname without a migration strategy and expect DNS clients to infer new microservice routes.
- Deploy one fixed EC2 server forever with the obsolete monolith because compatibility forbids any backend redesign.
Explanation
Correct Answer
B:A facade preserves externally consumed shapes while routing to new internals, allowing backend evolution without immediate breaking changes to clients.
Incorrect Answers
A:Exposing backend storage details breaks client contracts and expands coupling and security concerns.
C:DNS cannot automatically translate the semantics of an old business API into new resource and schema contracts.
D:Compatibility can be maintained with an adapter while internal implementations change under controlled testing.
Question 496
A backward-compatible version of a shopping API is ready for production. The business wants only 5% of users on the new version initially and automated rollback if error rates rise. What release approach fits?
- Run both API versions but never monitor their failure rates, assuming two versions make the release safe.
- Change the S3 storage class of API logs instead of controlling deployment traffic percentages.
- Use a supported canary or weighted rollout strategy with health and error metrics that trigger rollback.
- Switch 100% of production traffic immediately and remove the previous version before observing errors.
Explanation
Correct Answer
C:A small traffic slice limits exposure while testing real usage; reliable metrics and rollback thresholds protect the majority of users.
Incorrect Answers
A:Parallel deployment alone does not detect regressions or determine when traffic should be reverted.
B:Log storage cost settings do not distribute live requests between application versions.
D:A full irreversible cutover eliminates the gradual risk reduction and rollback behavior requested.
Question 497
A publisher changes an EventBridge event field from an integer to a nested object. Older consumers still expect the integer and fail when new messages arrive. Which architecture discipline prevents this kind of breakage?
- Apply event-schema compatibility and versioning rules with consumer-aware testing before publishing breaking changes.
- Increase the event bus throughput quota because schema incompatibility is only a capacity bottleneck.
- Assume EventBridge automatically converts every incompatible business field into each subscriber's expected data type.
- Remove all event consumers temporarily and delete the incompatible records without preserving business outcomes.
Explanation
Correct Answer
A:Event-driven systems are loosely coupled only when producers honor schema contracts or provide explicit versions and compatibility windows.
Incorrect Answers
B:Changing request rate limits does not make an older parser understand a structurally incompatible message.
C:Event routing does not guarantee arbitrary producer-to-consumer schema translation for custom payloads.
D:Dropping subscribers sacrifices processing and fails to establish a safe future schema evolution contract.
Question 498
A bursty Lambda API opens thousands of connections through RDS Proxy. Its developers put the database password in source code and plan to keep it unchanged for years. What security integration should be used?
- Give the Lambda function root-account access keys to avoid configuring database credentials and roles.
- Store database credentials in a supported Secrets Manager integration and scope proxy/client access appropriately.
- Disable TLS and database login controls between the proxy and RDS to maximize pool performance.
- Embed RDS master credentials into every Lambda deployment archive because pooling encrypts all application packages automatically.
Explanation
Correct Answer
B:Managed secret retrieval and rotation-aware design limit credential exposure while the proxy performs supported connection reuse.
Incorrect Answers
A:Root credentials increase blast radius and do not provide narrow managed secret handling for application database access.
C:Removing network encryption and authentication is not a safe prerequisite for supported proxy pooling.
D:Connection pooling does not protect hardcoded secrets copied into build artifacts and deployed function code.
Question 499
A payment intake service sometimes times out after sending an SQS FIFO message and retries the same logical submission. Consumers must receive no additional duplicate copy from immediate retries within the FIFO deduplication interval, but still must handle delayed repeated business requests safely. Which combined design is most appropriate?
- Switch to an SQS Standard queue, which guarantees exactly-once delivery with no duplicate messages.
- Increase the FIFO queue retention period because retention duration is used as the unique event deduplication identifier.
- Disable message-group identifiers so the FIFO queue merges duplicates from every customer regardless of order.
- Set a stable MessageDeduplicationId per logical submission for immediate retries and implement consumer idempotency for later duplicate business requests.
Explanation
Correct Answer
D:FIFO deduplication IDs suppress duplicate submissions within the deduplication window, but application idempotency remains needed for later repeats and end-to-end effects.
Incorrect Answers
A:SQS Standard provides at-least-once delivery and possible duplicates; it does not give the requested FIFO deduplication behavior.
B:Retention determines how long a message is stored, not identity of a logical message for FIFO deduplication.
C:FIFO messages require message-group semantics; omitting IDs does not create a safe deduplication scheme.
Question 500
A product-catalog API repeatedly queries the same 200 most popular SQL rows. The RDS writer has adequate connection headroom but high CPU from identical read queries; prices may be up to 30 seconds old, but updates must eventually be reflected. Which approach reduces repeated SQL execution?
- Increase RDS Proxy client connections and assume it caches SQL result sets for repeat reads.
- Replace the catalog read endpoint with an RDS automated snapshot, which remains an interactive SQL connection to current prices.
- Add an ElastiCache Redis-compatible cache-aside layer keyed by product ID with a short TTL and update-aware invalidation.
- Turn on S3 Object Lock on an unrelated bucket so SQL SELECT requests are served from immutable objects.
Explanation
Correct Answer
C:Cache-aside reads from the cache first and populates missing entries from RDS; TTL and invalidation enforce the allowed staleness window.
Incorrect Answers
A:RDS Proxy pools connections and manages failover; it does not provide a SQL result cache for identical catalog reads.
B:Automated snapshots are backups, not a low-latency query endpoint serving live relational rows.
D:Object Lock protects S3 object retention and cannot intercept an RDS SQL API for caching.
