AWS SAA-C03 Design Resilient Architectures Practice Questions

Free SAA-C03 Design Resilient Architectures practice test questions

8 questions from this domain with answers and explanations - different from the samples on the main AWS SAA-C03 page. Sign up free to practice the full set.

1. Question 1Design Resilient Architectures

   An e-commerce company runs its product catalog on an Amazon Aurora MySQL cluster with one writer instance and two reader instances across two Availability Zones. During a peak sales event, the writer instance fails. The company requires the database to recover automatically with minimal downtime. What happens when the Aurora writer instance fails?

   • AAurora promotes one reader to the new writer and updates the cluster endpoint DNS, typically finishing failover in under 30 secondsCorrect
   • BAurora keeps the entire cluster read-only and waits for an administrator to manually select and promote one of the two reader instances to become the new writer
   • CAurora provisions a brand-new writer instance from the most recent cross-Region backup and then repoints the cluster endpoint to that freshly restored instance
   • DAurora rebuilds the failed writer from the latest automated snapshot in place, restoring write capability in roughly 15 to 30 minutes while the readers stay online
   ✓ Correct answer: A

   Aurora is designed with automatic failover capability where if the writer instance fails, Aurora automatically promotes one of the available reader replicas to serve as the new writer. The cluster endpoint DNS name automatically points to the new writer, allowing applications to reconnect without code changes. This automatic promotion typically completes in under 30 seconds, meeting the requirement for minimal downtime.

   Why the other options are wrong

   • BAurora failover is automatic when reader instances exist; it promotes a replica without any administrator action, so the cluster does not stay stuck in read-only awaiting a manual promotion.
   • CAurora shares a single distributed storage layer across the cluster, so failover simply promotes an existing in-Region replica rather than restoring a cross-Region backup, which would be far slower and lose recent writes.
   • DBecause the storage volume survives the instance failure, Aurora promotes a running replica in seconds rather than rebuilding from a snapshot, so the 15 to 30 minute in-place restore does not occur.
2. Question 2Design Resilient Architectures

   A company has an application that writes data to an Amazon EBS volume. The company needs to create a backup strategy that takes consistent application-level backups of the EBS volume every 12 hours and retains them for 30 days. The solution should require minimal operational effort. Which service should the solutions architect use?

   • AManually create EBS snapshots twice daily through the AWS Management Console.
   • BWrite a cron job on the EC2 instance to call the create-snapshot API every 12 hours.
   • CUse AWS Backup to create a backup plan with a 12-hour schedule and 30-day retention.Correct
   • DUse Amazon Data Lifecycle Manager (DLM) to create a lifecycle policy for automated EBS snapshots with a 12-hour schedule and 30-day retention.
   ✓ Correct answer: C

   AWS Backup provides centralized backup management with support for scheduled backups and retention policies. Creating a backup plan with a 12-hour schedule ensures snapshots every 12 hours, and setting retention to 30 days automatically manages snapshot cleanup. This requires minimal operational effort compared to manual or custom scripted approaches.

   Why the other options are wrong

   • AManually create EBS snapshots twice daily through the AWS Management Console is incorrect because this requires manual effort and is not scalable.
   • BWrite a cron job on the EC2 instance to call the create-snapshot API every 12 hours is incorrect because this requires maintaining scripts on each instance.
   • DUse Amazon Data Lifecycle Manager (DLM) to create a lifecycle policy for automated EBS snapshots with a 12-hour schedule and 30-day retention is incorrect; while DLM works, AWS Backup is the more comprehensive solution.
3. Question 3Design Resilient Architectures

   FrostByte Gaming hosts a multiplayer game backend on AWS. The application stores player profile data in Amazon DynamoDB. The architects need to ensure that the DynamoDB table can handle sudden, unpredictable spikes in read and write traffic without throttling. Which capacity mode should they choose?

   • AOn-demand capacity mode.Correct
   • BProvisioned capacity mode with reserved capacity.
   • CProvisioned capacity mode with Auto Scaling enabled.
   • DProvisioned capacity mode with a fixed high throughput allocation.
   ✓ Correct answer: A

   On-demand capacity mode automatically scales DynamoDB throughput up and down based on actual traffic without requiring capacity planning. For workloads with sudden, unpredictable traffic spikes, on-demand mode ensures the table can handle traffic bursts without throttling. You pay for the consumed capacity rather than provisioning capacity upfront.

   Why the other options are wrong

   • BProvisioned capacity mode with reserved capacity is incorrect because reserved capacity cannot dynamically scale for unpredictable spikes.
   • CProvisioned capacity mode with Auto Scaling enabled is incorrect because Auto Scaling has a delay before adding capacity; sudden spikes might still cause throttling.
   • DProvisioned capacity mode with a fixed high throughput allocation is incorrect because you'd pay for unused capacity during low-traffic periods.
4. Question 4Design Resilient Architectures

   SpectraLabs runs batch processing jobs that analyze genomic data. The jobs can take up to 4 hours to complete. If a job fails partway through, it must restart from the beginning. The company wants to minimize cost while ensuring the jobs complete reliably. Which compute option is MOST appropriate?

   • AUse Amazon EC2 On-Demand Instances to guarantee the jobs run to completion without any interruption risk.
   • BUse AWS Lambda functions and restart the analysis if it exceeds the 15-minute execution limit.
   • CUse EC2 Spot Instances with interruption handling that checkpoints progress to S3.Correct
   • DUse Amazon EC2 Spot Instances with no interruption handling for the lowest possible price.
   ✓ Correct answer: C

   EC2 Spot Instances can reduce compute costs by up to 90% compared to On-Demand pricing, making them ideal for cost-sensitive batch workloads. By implementing Spot interruption handling - which gives a 2-minute warning before termination - jobs can checkpoint their current progress to S3. On restart after interruption, the job can resume from the last checkpoint rather than from scratch, combining cost efficiency with job reliability for long-running genomic analysis tasks.

   Why the other options are wrong

   • AOn-Demand guarantees availability but at full price with no discount, so it does not minimize cost, which is a stated goal for this interruption-tolerant workload.
   • BLambda has a hard 15-minute timeout, so it cannot run a 4-hour genomic analysis, and repeatedly restarting after the limit would never let the job finish.
   • DSpot without interruption handling means a reclaimed instance loses all in-flight work and the job restarts from the beginning every time, undermining reliable completion despite the low price.
5. Question 5Design Resilient Architectures

   What is the purpose of Amazon ElastiCache?

   • ADurable block storage volumes attached to EC2 instances
   • BScalable object storage for files and backups
   • CIn-memory caching to improve application performanceCorrect
   • DFully managed message queuing between services
   ✓ Correct answer: C

   Amazon ElastiCache is a fully managed in-memory caching service that supports Redis and Memcached engines. It sits in front of databases or backend services and stores frequently accessed data in RAM, reducing latency from milliseconds to microseconds and relieving pressure on the primary data store for read-heavy workloads.

   Why the other options are wrong

   • ABlock storage attached to EC2 is the role of Amazon EBS, not ElastiCache.
   • BObject storage for files and backups is provided by Amazon S3, not ElastiCache.
   • DMessage queuing between services is provided by Amazon SQS, not ElastiCache.
6. Question 6Design Resilient Architectures

   A company hosts a static marketing website in an Amazon S3 bucket served through Amazon CloudFront. The team wants the site to remain available to users even if the S3 origin in the primary Region becomes unavailable. Which approach provides automatic failover to a secondary origin?

   • AEnable S3 Cross-Region Replication and update the bucket policy
   • BConfigure a CloudFront origin group with a primary and secondary originCorrect
   • CAdd a Route 53 latency-based routing record for both buckets
   • DEnable S3 Transfer Acceleration on both buckets
   ✓ Correct answer: B

   CloudFront origin groups let you designate a primary origin and a failover origin. When CloudFront receives configured failure status codes (such as 5xx) or times out from the primary, it automatically retries the request against the secondary origin, keeping the site available if the primary S3 origin fails.

   Why the other options are wrong

   • ACross-Region Replication keeps a copy of objects in another bucket but does not by itself redirect CloudFront traffic when the primary origin fails.
   • CLatency-based routing optimizes for performance among healthy endpoints but is not configured on the CloudFront origin layer to fail over S3 origins.
   • DTransfer Acceleration speeds uploads/downloads over the AWS edge network; it provides no failover capability.
7. Question 7Design Resilient Architectures

   A company uses Amazon Aurora MySQL for a read-heavy reporting application. Reporting queries are degrading the performance of the primary writer. The team wants to scale reads and improve resiliency with automatic failover. Which approach is BEST?

   • AAdd Aurora Replicas and route reporting queries to the reader endpointCorrect
   • BEnable the Aurora Multi-Master write option on every instance in the cluster
   • CTake more frequent automated snapshots of the Aurora cluster storage
   • DIncrease the writer instance to the largest available instance class
   ✓ Correct answer: A

   Directing reporting queries to the reader endpoint offloads them from the writer, and Aurora can automatically promote a replica if the writer fails, improving both performance and resiliency. Replicas share the cluster storage volume so replication lag is minimal.

   Why the other options are wrong

   • BMulti-Master is for multiple write nodes and adds conflict-resolution complexity; it does not address offloading read reporting queries.
   • CMore frequent snapshots improve recovery points but do nothing to offload live read load or improve failover.
   • DVertical scaling of the writer is a temporary fix that does not separate read and write workloads and hits a hard ceiling.
8. Question 8Design Resilient ArchitecturesSelect all that apply

   A solutions architect is designing a highly available, fault-tolerant three-tier web application in a single AWS Region. Which TWO design choices improve resilience against an Availability Zone failure? (Choose two.)

   • ADeploy the EC2 web tier in an Auto Scaling group spanning at least two Availability ZonesCorrect
   • BUse an Amazon RDS Multi-AZ deployment for the database tierCorrect
   • CPlace all instances in a single AZ within a cluster placement group for low latency and simplified network configuration.
   • DStore session state on local instance store volumes
   ✓ Correct answer: A, B

   An Auto Scaling group across multiple AZs keeps the web tier serving traffic if one AZ fails, while RDS Multi-AZ provides a synchronous standby that fails over automatically. Together they remove single-AZ points of failure across compute and database tiers.

   Why the other options are wrong

   • CA cluster placement group concentrates instances in one AZ, increasing risk if that AZ fails.
   • DLocal instance store is ephemeral and tied to one instance/AZ, so session data would be lost during an AZ failure.

How Design Resilient Architectures is tested

This domain holds 176 of the 675 questions in the SAA-C03 bank, about 26%. The mix is 150 single-answer multiple choice, 24 multiple-response, 1 yes/no scenario and 1 ordering, so it is worth practising the formats as well as the content.

Once you have a few attempts recorded, CertGrid scores every domain separately and points you at the weakest one, so you can drill Design Resilient Architectures on its own rather than re-running full-length mocks.

Other SAA-C03 exam domains

• Design Secure Architectures202 questions
• Design High-Performing Architectures159 questions
• Design Cost-Optimized Architectures138 questions
• All SAA-C03 practice questions675 total
• Design Resilient Architectures study notesKey concepts
• AWS practice examsAll AWS

SAA-C03 Design Resilient Architectures FAQ