Amazon AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 30

Question 290

A cache-aside service writes a new product price to DynamoDB but continues returning yesterday’s value from ElastiCache until the TTL expires. Which data-consistency step is missing?

Increase ElastiCache key TTL to avoid inconsistency.
Add more EC2 worker threads without changing cache invalidation.
Turn up ElastiCache read replica count without invalidating the old price key that every cache replica still contains.
Invalidate or update the corresponding cache key when the authoritative value changes.

Explanation

Correct Answer

Cache-aside reads can serve stale entries until expiration unless writes trigger targeted invalidation or refresh.

Incorrect Answers

Longer TTL increases how long obsolete values can persist.

Concurrency does not correct a stale cache item.

Additional cache replicas can increase read capacity but do not make stale entries current. Coordinate write-through or targeted invalidation with authoritative data changes.

Question 291

A web function opens a new RDS database connection on every invocation. During a sudden Lambda surge, RDS rejects connections although CPU is low. What architecture gap explains it?

Enable an RDS read endpoint for the same writer but keep the Lambda handler opening new writer sessions per invocation, expecting reader routing to pool primary connections automatically.
Turning on S3 Intelligent-Tiering raises the RDS connection limit.
Increasing query page sizes necessarily eliminates connection-count exhaustion.
The application lacks effective connection pooling or a database proxy that absorbs bursty client connections.

Explanation

Correct Answer

RDS Proxy reuses database connections and can reduce overload from many ephemeral compute clients.

Incorrect Answers

A reader endpoint routes supported read traffic; it does not pool new write connections to the writer. RDS Proxy (and appropriate pooling/reuse) addresses connection churn.

Object storage class has no effect on DB connection slots.

Per-query row count does not govern number of sessions opened.

Question 292

A website places personalized account pages behind CloudFront with a cache policy that ignores session context. A user receives an HTML response rendered for another user. What is the failure?

CloudFront signed URLs automatically encrypt every page body for individual users.
ALB target health checks are required for per-user cache confidentiality.
The cache key does not distinguish personalized responses; private content is being cached and reused across identities.
Add an RDS reader replica in the origin VPC but continue caching a user-specific HTML representation under one shared CDN key for all identities.

Explanation

Correct Answer

Cache only safe shared responses or correctly configure cache keys/origin headers and authorization for user-specific content.

Incorrect Answers

A signed URL is an access-control mechanism, not a substitute for cache-key isolation.

Health checking does not determine whether cache entries are user-specific.

Database read distribution has no bearing on CloudFront cache-key identity isolation. Personalized pages should not be reused across users without correct cache policies and authorization.

Question 293

A DynamoDB stream consumer has been disabled for longer than the stream retention period. When restarted, it cannot read the missed early records. Why?

DynamoDB always stores Streams changes forever when PITR is enabled.
A high Route 53 TTL causes stream records to be deleted as soon as they are written.
Enable DynamoDB PITR after the outage and assume this reconstructs every already-expired individual Streams change event in original sequence.
DynamoDB Streams keeps change records for a bounded retention period, after which expired records are unavailable for replay.

Explanation

Correct Answer

A stream is not an indefinite durable audit archive; design recovery or export for longer outages.

Incorrect Answers

Point-in-time recovery of table state is not infinite stream event retention.

DNS has no control over stream retention.

PITR can restore table state within its configured window, but it is not an indefinite Streams event log. Expired change records are not replayed by enabling PITR.

Question 294

Two Kinesis consumers independently process every shard record. The second needs predictable low-latency delivery without competing for the first consumer’s shared get-record bandwidth. What is missing?

Increase S3 bucket size to change a Kinesis shard read limit.
A supported enhanced fan-out registered consumer and SubscribeToShard path for independently allocated throughput.
Give Kinesis consumers distinct KCL application names but leave them on the shared GetRecords polling path, assuming an application name alone assigns each dedicated shard bandwidth.
Decrease the number of shard partitions to improve every consumer’s throughput.

Explanation

Correct Answer

EFO consumers receive dedicated per-shard read throughput rather than sharing the normal polling quota.

Incorrect Answers

S3 capacity does not alter Kinesis read throughput.

Separate application checkpoints do not reserve enhanced fan-out throughput. Register consumers and use SubscribeToShard for eligible dedicated read throughput.

Fewer shards can worsen available read capacity.

Question 295

A legacy application expects AMQP/JMS semantics and transactions but the architects substituted an SQS queue without changing client protocol support. Why can it not connect?

SQS is an AWS API queue rather than a drop-in AMQP/JMS broker; use a compatible managed broker or adapt the client.
SQS can be reached as a native AMQP endpoint without any adapter.
Translate only the JMS destination hostname to a Route 53 CNAME for an SQS endpoint, expecting DNS to translate the application JMS framing and acknowledgement protocol automatically.
S3 event notifications guarantee transactional JMS queue semantics.

Explanation

Correct Answer

Amazon MQ provides managed broker compatibility for supported protocols; switching transport requires explicit client integration.

Incorrect Answers

SQS interfaces are AWS APIs and SDKs, not an AMQP/JMS wire-compatible broker.

DNS indirection cannot change message wire protocols or transaction semantics. A compatible broker or explicit client rewrite is required.

Object event notifications do not implement a generic JMS transaction broker.

Question 296

An S3 bucket notification targets a Lambda function, but the upload key contains a prefix excluded by the notification’s filter. The object exists and Lambda permissions are valid. Why no invocation?

The configured event notification prefix/suffix filter does not match the uploaded object key.
S3 notifications always deliver for every prefix regardless of configuration.
Enable the S3 object-created event type but leave its key prefix filter set to incoming/ while upload keys begin with raw/, assuming event type selection overrides prefix filters.
The bucket must be public-read to emit ObjectCreated notifications.

Explanation

Correct Answer

S3 only emits matching configured event notifications to the destination; test the exact event type and key filters.

Incorrect Answers

Filters deliberately restrict eligible object keys.

Event type and key filters both must match. An ObjectCreated event on an excluded key will not invoke the target regardless of Lambda permission correctness.

Event delivery does not require making the source object public.

Question 297

An AWS Batch job queue is enabled with a managed EC2 compute environment. A new analytics job requests 32 vCPUs in one container, but the environment is capped at maxvCpus=16 and no larger compatible placement is allowed. The job remains RUNNABLE. What capacity mismatch blocks placement?

Allow the job to use a smaller supported vCPU requirement without verifying that its application fits, assuming any smaller number automatically preserves successful execution.
Increase the priority of the same Batch job queue, assuming priority allows a single 32-vCPU job to fit a compute environment capped at 16 vCPUs.
The job’s 32-vCPU request exceeds every eligible capacity placement under the environment maximum and instance selection; use an eligible larger environment or a legitimately smaller job.
Adjust an SQS FIFO queue MessageGroupId for the jobs and assume SQS group settings expand the AWS Batch compute environment maximum vCPU capacity.

Explanation

Correct Answer

A job cannot be placed when its resource requirements exceed every allowed compute placement, even though the job queue is otherwise enabled. Align maximum vCPUs, valid instance types and per-job requests while checking quotas and actual capacity.

Incorrect Answers

Reducing job vCPU requests can make placement feasible only if the workload is valid with less compute. It must be tested and correspond to an eligible instance/compute environment rather than altered blindly.

Queue priority influences scheduling order among eligible jobs, not an overriding exemption from the environment maximum and compatible instance shapes.

SQS ordering identifiers do not allocate AWS Batch vCPUs or fix a job request exceeding compute environment limits.

Question 298

A workload sends many messages in bursts. CloudWatch shows the SQS standard queue has a growing ApproximateNumberOfMessagesVisible, but producers report successful SendMessage. Where is the bottleneck?

Successful SendMessage means consumers have already committed all side effects.
The SQS queue has become an immutable S3 archive rather than a work queue.
Consumers are processing/acknowledging messages more slowly than producers add new ones.
Increase only producer SendMessage concurrency while leaving the same slow worker processing rate and assume a successful send means the backlog is draining.

Explanation

Correct Answer

The visible backlog metric indicates queued work awaiting receipt; inspect consumer fleet, processing rate and errors.

Incorrect Answers

Queue acceptance is not consumer completion.

A backlog is expected queue behavior, not a storage-class conversion.

Successful SQS sends show producer acceptance, not consumer completion. More producer throughput can make the backlog grow faster.

Question 299

ECS Service Connect allows clients to discover a named service but requests fail after a deployment because the ECS service’s published port alias no longer matches the listening container port. What failed?

Changing a DynamoDB GSI adds the missing TCP listener.
The configured Service Connect endpoint/port mapping no longer corresponds to the application listener.
Service Connect always opens every container port regardless of task definitions.
Enable ECS Service Connect client aliases with a different advertised port but leave the server task definition’s application listener on the old port, assuming aliases rewrite task listeners.

Explanation

Correct Answer

The mesh/service discovery layer still needs a valid target port and task definition network setup.

Incorrect Answers

Database indexes are unrelated to container network endpoints.

Service registration does not infer an arbitrary process listener.

Service discovery names/aliases do not silently reconfigure the container process listener. Match service configuration, task port mapping and application listener.

Question 300

A Cloud Map service’s DNS record still returns an instance address that was terminated hours ago because the service instance was not deregistered and its health state is stale. What is wrong?

Service registry membership and health state were not updated when the instance was removed.
A CloudWatch alarm on S3 PUT requests directly removes Cloud Map entries.
A more expensive EC2 instance type changes a stale registry’s DNS result.
Lower Cloud Map DNS TTL to five seconds but leave the terminated instance registered as healthy in the registry, assuming TTL deletes the underlying record itself.

Explanation

Correct Answer

Cloud Map discovery requires correct registration/deregistration and supported health checks to avoid stale endpoints.

Incorrect Answers

Storage alarm metrics do not synchronize service registry membership.

Capacity alone does not update discovery records.

Short TTL reduces how long a DNS response may be cached but cannot automatically deregister a stale instance. Registration and health must reflect current running endpoints.

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

Question 290

Explanation

Correct Answer

Incorrect Answers

Question 291

Explanation

Correct Answer

Incorrect Answers

Question 292

Explanation

Correct Answer

Incorrect Answers

Question 293

Explanation

Correct Answer

Incorrect Answers

Question 294

Explanation

Correct Answer

Incorrect Answers

Question 295

Explanation

Correct Answer

Incorrect Answers

Question 296

Explanation

Correct Answer

Incorrect Answers

Question 297

Explanation

Correct Answer

Incorrect Answers

Question 298

Explanation

Correct Answer

Incorrect Answers

Question 299

Explanation

Correct Answer

Incorrect Answers

Question 300

Explanation

Correct Answer

Incorrect Answers

AWS Certified Solutions Architect - Associate SAA-C03 Practice Tests