Home AWS SAA-C03: Solutions Architect Associate SAA-C03 Questions 71-80: Data Services and Application Resilience

SAA-C03 Questions 71-80: Data Services and Application Resilience

0
15
SAA-C03 data services questions, Set 8: Data Services and Application Resilience, practice questions 71-80

The AWS Certified Solutions Architect – Associate (SAA-C03) exam covers resilient architectures, worth 26% of scored content, and high-performing architectures, worth 24%. These domains test how applications recover from failures, distribute work, and choose data services that match access patterns. These SAA-C03 data services questions cover messaging, container recovery, database design, and data ingestion.

This is the eighth set in our AWS Certified Solutions Architect – Associate SAA-C03 series (Questions 71-80). It follows SAA-C03 Questions 61-70 on networking and hybrid connectivity. Every question is an original practice scenario mapped to the official exam objectives, not a copy of a real exam item.

Before choosing a service, identify whether each event needs one worker or several independent subscribers. For database questions, separate how data is distributed from how recently written data becomes visible.

Question 71: Each Team Needs Every Event

A ticketing platform publishes reservation events. Billing, notification, and analytics workers must each receive every event and maintain separate backlogs during short outages. The publisher should send each event to one destination, without maintaining delivery logic for each team. Ordering is not required. Which design best meets these requirements?

A) Send events to one SQS queue that all three teams poll.

B) Publish to one SNS topic with direct HTTP subscriptions for the workers.

C) Publish to one SNS topic with a separate SQS subscription per team.

D) Send events to three SQS queues using delivery logic in the publisher.

👁 Reveal Answer

Correct Answer: C

Explanation: SNS fans out a published event to the subscribed SQS queues. Each team then consumes its own queue at its own pace, preserving an independent backlog while its workers are unavailable. A creates competing consumers rather than a separate event copy for each team. B does not provide the requested per-team SQS backlogs. D can create separate backlogs, but requires the publisher to manage three deliveries instead of publishing once.

Reference: Fanout Amazon SNS notifications to Amazon SQS queues for asynchronous processing

Question 72: Parallel Work Without Reordering

A fulfillment application sends updates to one SQS FIFO queue in the correct sequence for each shipment. Workers must preserve that sequence within a shipment but process unrelated shipments concurrently. All messages currently use the same message group ID, limiting concurrency. Which change meets both requirements?

A) Use the shipment ID as the message group ID for its updates.

B) Use a new random message group ID for every individual update.

C) Keep the shared message group ID and add more polling workers.

D) Use a standard queue with the shipment ID as the message group ID.

👁 Reveal Answer

Correct Answer: A

Explanation: A FIFO queue preserves order within a message group while allowing different groups to be consumed concurrently. Using one group per shipment keeps related updates together without serializing all shipments. B separates related updates into different groups, losing the required ordering relationship. C leaves all work in one ordered group, so additional workers do not remove that constraint. D uses a standard queue, where message group IDs do not provide FIFO ordering.

Reference: Using the message group ID with Amazon SQS FIFO Queues

Question 73: Recovering a Failed Workflow Step

A Step Functions workflow calls a Lambda function to reserve inventory. The task sometimes returns a named transient error, and the function safely handles repeated requests. The team wants a limited number of retries with increasing delays, followed by a manual-review state if that error persists. Which configuration requires the least custom orchestration code?

A) Add a matching Catch that routes directly to the manual-review state.

B) Increase the task timeout and add a Catch leading to manual review.

C) Add a matching Retry with backoff, but leave the error without a Catch.

D) Add a matching Retry with backoff and a Catch leading to manual review.

👁 Reveal Answer

Correct Answer: D

Explanation: Step Functions applies a matching retrier before a matching catcher. Configure a finite retry count and increasing intervals, then use Catch to transition to manual review after retries are exhausted. A sends the error to review immediately instead of retrying. B changes how long a task may run, not how a returned error is retried. C supplies retries but leaves the workflow without the required fallback transition when the error persists.

Reference: Handling errors in Step Functions workflows

Question 74: Balancing Availability and Host Utilization

An ECS service uses the EC2 launch type. Eligible container instances have sufficient capacity in three Availability Zones. The architect wants task placement to prioritize an even distribution across zones, then minimize unused memory on hosts within each zone. Which ordered placement-strategy list should the service use?

A) First binpack on memory, then spread on attribute:ecs.availability-zone.

B) First spread on attribute:ecs.availability-zone, then binpack on memory.

C) First spread on instanceId, then binpack on cpu.

D) First random placement, then spread on instanceId.

👁 Reveal Answer

Correct Answer: B

Explanation: ECS applies multiple placement strategies in their specified order. B first spreads the service’s tasks across Availability Zones, then uses memory bin packing within that distribution. Strategies are best effort rather than absolute placement guarantees. A reverses the required priorities. C spreads across hosts rather than explicitly balancing zones, and optimizes CPU instead of memory. D introduces random placement and never applies the required memory-packing strategy.

Reference: Use strategies to define Amazon ECS task placement

Question 75: Keeping the Worker Fleet Running

A team launches four long-running stateless workers as standalone ECS tasks on Fargate. When a task stops, an operator launches its replacement. The team also wants tasks that fail configured container health checks to be replaced automatically. Which change provides this behavior with the least operational effort?

A) Launch the standalone tasks hourly through EventBridge Scheduler.

B) Run the task definition in an ECS service with a desired count of four.

C) Register the standalone tasks with a load balancer and monitor target health.

D) Configure a CloudWatch alarm to notify operators when a task stops.

👁 Reveal Answer

Correct Answer: B

Explanation: An ECS service maintains the desired number of tasks and replaces stopped or unhealthy tasks through its service scheduler. Replacement timing still depends on capacity and service settings. A launches tasks on a schedule rather than maintaining the desired healthy population. C adds traffic distribution and health monitoring but does not give standalone tasks a service scheduler. D detects a failure but still requires the manual replacement process the team wants to eliminate.

Reference: Amazon ECS services

Question 76: Choosing a Key for a Busy Workload

A new DynamoDB table will store inspection readings from thousands of machines. Each reading has one of three inspection status values. Each machine sends readings at approximately the same rate, and applications retrieve recent readings for one machine at a time. The team wants to distribute writes while supporting that access pattern. Which primary-key design is best?

A) Partition key: inspection status; sort key: machine ID and reading time.

B) Partition key: current date; sort key: machine ID and reading time.

C) Partition key: a constant fleet label; sort key: machine ID and reading time.

D) Partition key: machine ID; sort key: reading time and unique reading ID.

👁 Reveal Answer

Correct Answer: D

Explanation: Machine IDs provide many actively used partition-key values because the machines send readings at similar rates. The sort key supports time-based retrieval within one machine’s data. A concentrates writes into a small number of status values. B directs the current day’s writes to the same partition-key value. C puts every reading under one partition-key value. Distinct sort keys do not substitute for a partition-key design that distributes the workload.

Reference: Designing partition keys to distribute your workload in DynamoDB

Question 77: Showing a Confirmed Update Immediately

An application updates a customer preference in a single-Region DynamoDB table and receives a successful response. No other writer changes that item. The confirmation page knows the complete primary key but sometimes displays the old value. There is no application cache. Which read approach ensures that the page reflects the completed write?

A) Read the base-table item with GetItem and ConsistentRead=true.

B) Read the base-table item with GetItem and ConsistentRead=false.

C) Query a global secondary index with ConsistentRead=true.

D) Wait briefly, then read with GetItem and omit ConsistentRead.

👁 Reveal Answer

Correct Answer: A

Explanation: A strongly consistent read of the base table reflects prior successful writes. The known primary key lets the application retrieve the item directly without an index lookup. B explicitly requests eventual consistency, so a recent update might not yet appear. C requests a consistency mode that global secondary indexes do not support. D still uses eventual consistency by default; an arbitrary brief delay does not provide a strong-read guarantee.

Reference: DynamoDB read consistency

Question 78: Following Connections Across Records

A fraud investigation application stores accounts, devices, and payment relationships. Analysts need interactive queries that traverse several relationships to find connected accounts. The team wants a managed database with native Gremlin support rather than implementing graph traversal in application code. Which database should it select?

A) Amazon Aurora PostgreSQL for relational tables and SQL queries.

B) Amazon DynamoDB for partition-key and sort-key access patterns.

C) Amazon Neptune for property graphs and native Gremlin queries.

D) Amazon Redshift for warehouse tables and SQL analytical queries.

👁 Reveal Answer

Correct Answer: C

Explanation: Neptune is a managed graph database designed for highly connected datasets and supports Gremlin for property-graph traversal. That matches both the relationship-heavy workload and the required query interface. A can represent relationships in tables but does not supply the requested native Gremlin graph engine. B would require a different access model and application traversal logic. D targets SQL analytics rather than the required native Gremlin interface for interactive relationship traversal.

Reference: What Is Amazon Neptune?

Question 79: Moving Files Into Cloud Processing

A research lab stores completed instrument files on an existing on-premises NFS server. Every night, it must copy those files to Amazon S3 for analysis, with transfer verification and minimal custom code. Network capacity is sufficient. It does not need a new file-access interface for users. Which approach best fits?

A) Write a scheduled transfer program that copies files and compares checksums.

B) Replace the file workflow with producers that send records to Data Firehose.

C) Schedule a file-copy script and manually verify the destination files.

D) Configure a scheduled DataSync task from NFS to S3 with verification.

👁 Reveal Answer

Correct Answer: D

Explanation: DataSync transfers file data from NFS storage to S3 and provides task options for scheduling and data verification. It fits recurring transfers of existing files without creating a custom transfer application. A could implement the workflow but adds development and maintenance. B changes a file-transfer requirement into a streaming-record ingestion design. C still requires a custom script and manual verification rather than the managed transfer and verification process.

Reference: Creating a task for transferring your data

Question 80: Delivering Telemetry Without a Consumer Fleet

A company can configure its applications to send telemetry records directly to an AWS ingestion service. It only needs buffered delivery into an S3 bucket; it does not need independent consumers or replay from the ingestion layer. The team wants to avoid writing a delivery application or managing servers. Which design is the best fit?

A) Send records to Amazon Data Firehose with an S3 destination.

B) Send records to Kinesis Data Streams with a custom S3 consumer.

C) Send records to SQS with an EC2 worker that writes S3 objects.

D) Send records to an HTTP application on EC2 that batches S3 writes.

👁 Reveal Answer

Correct Answer: A

Explanation: Amazon Data Firehose accepts streaming records, buffers them, and delivers them to a configured destination such as S3. It removes the need to build a delivery consumer for this requirement. B introduces custom consumer code even though replay and independent consumers are unnecessary. C requires operating an EC2 worker and implementing delivery logic. D likewise adds an application and servers for a function that Firehose already manages.

Reference: What is Amazon Data Firehose?

Study Tips for Data Services and Application Resilience

  • Separate delivery from processing. An SNS topic can fan out to independent SQS backlogs, while a FIFO message group defines the scope of ordering.
  • Prefer managed recovery. Step Functions retries handle transient task errors, and ECS services maintain a desired task population.
  • Match keys to activity. Many evenly active partition keys distribute DynamoDB work, while strong reads address visibility of completed writes.
  • Identify the input shape. DataSync moves existing files between storage locations; Data Firehose delivers incoming streaming records.

Keep Practicing for AWS Solutions Architect Associate

Review SAA-C03 Questions 61-70 for networking decisions, then revisit these SAA-C03 data services questions to compare messaging and database choices. The next set will cover storage security and cost controls.

Cloud Tech
Cloud Tech is a small, independent team that writes free practice questions for Microsoft Azure, AWS and CompTIA certification exams. We write every question ourselves against the current official exam outline, explain why each option is right or wrong, link to the vendor's documentation, and update or archive sets when an exam changes. Spotted a mistake? Email [email protected].