Designing resilient architectures is worth 26% of SAA-C03. These questions ask how a workload keeps running when an instance, an Availability Zone, or a whole Region fails, and how components are decoupled so that a spike or a bad message does not take everything down. The key is to match each design to the recovery time objective (RTO) and recovery point objective (RPO) in the scenario.
This is the fourth set in our SAA-C03 series (Questions 31-40). It follows SAA-C03 Questions 21-30 on data protection and network security. Every question is an original practice scenario mapped to the official exam guide, not a copy of a real exam item.
Look for the words that set the requirement, such as “minutes”, “lowest cost”, or “no data loss”, before you choose.
Question 31: What Happens When the Primary Database Fails
A company runs Amazon RDS for PostgreSQL as a Multi-AZ DB instance deployment. The primary instance fails during the night. What happens to the application, which connects using the DB instance’s endpoint?
A) An administrator must change the application’s connection string to the standby’s address
B) A read replica is promoted, and all recent writes are lost
C) The database stays unavailable until AWS Support restores it from a snapshot
D) RDS automatically fails over to the synchronous standby in another Availability Zone and points the same endpoint to it, usually within one to two minutes
👁 Reveal Answer
Correct Answer: D
Explanation: In a Multi-AZ DB instance deployment, RDS keeps a synchronously replicated standby in a different Availability Zone. If the primary fails, RDS promotes the standby and updates the DNS record behind the same endpoint, so the application reconnects without configuration changes. Failover typically completes in 60-120 seconds. A defeats the purpose of Multi-AZ. B confuses Multi-AZ with read replicas, which use asynchronous replication and are not promoted automatically. C describes having no high availability at all.
Question 32: Users Logged Out When Instances Scale In
A web application stores user session data in each EC2 instance’s memory. When Auto Scaling removes instances, users connected to them are logged out and lose their shopping carts. What change makes the web tier stateless and resilient?
A) Turn on sticky sessions on the load balancer and never scale in
B) Store session data in a shared store such as Amazon ElastiCache or Amazon DynamoDB
C) Use larger instances so fewer are needed
D) Save session data to each instance’s EBS volume
👁 Reveal Answer
Correct Answer: B
Explanation: Moving session state out of the instances into a shared, highly available store makes every instance interchangeable. Any instance can serve any user, so scaling in or losing an instance no longer logs users out. ElastiCache offers in-memory speed, and DynamoDB offers fully managed durability. A ties users to specific instances and gives up elasticity. C reduces the number of instances but not the problem. D still keeps state on individual instances.
Reference: Amazon ElastiCache (Choosing an AWS NoSQL database)
Question 33: One File System for Many Instances
A content management system runs on Linux EC2 instances in an Auto Scaling group across three Availability Zones. All instances must read and write the same uploaded files at the same time, and the storage must grow automatically. What should the architect use?
A) One EBS volume attached to all instances
B) Instance store volumes with a nightly sync script
C) Amazon EFS, mounted on every instance
D) Amazon S3 Glacier Deep Archive
👁 Reveal Answer
Correct Answer: C
Explanation: Amazon EFS is a fully managed, elastic NFS file system that many EC2 instances can mount at the same time, including instances in different Availability Zones, and it grows and shrinks automatically as files are added and removed. A is limited: an EBS volume lives in one Availability Zone, and Multi-Attach is only for specific volume types and use cases. B loses data when instances stop and is never consistent. D is for archives, with retrieval times far too slow for live files.
Reference: What is Amazon Elastic File System?
Question 34: Orders Lost During a Traffic Spike
A front-end service sends orders directly to a processing service. During flash sales, the processing service is overwhelmed and orders are lost. How should the architect redesign this?
A) Put an Amazon SQS queue between the services so orders are stored durably and processed at the rate the workers can handle
B) Increase the timeout on the front-end service
C) Run the processing service on a single larger instance
D) Send orders by email to the warehouse
👁 Reveal Answer
Correct Answer: A
Explanation: Amazon SQS decouples the two services. The front end writes orders to the queue and returns quickly, the queue stores messages durably, and workers read at their own pace, scaling out (for example, with an Auto Scaling group based on queue depth) when the backlog grows. B only makes users wait longer before the failure. C still has a hard limit and creates a single point of failure. D is not an architecture.
Reference: AWS SAA-C03 exam guide, Task 2.1 (loosely coupled architectures)
Question 35: One Bad Message Blocks the Queue
Workers process messages from an SQS queue. A few malformed messages fail every time, return to the queue, and are retried endlessly, wasting capacity and hiding real problems. What should the team configure?
A) A longer visibility timeout so failures happen less often
B) A dead-letter queue with a redrive policy that moves a message after it has been received a set number of times
C) Delete the queue and recreate it every day
D) A FIFO queue so messages are processed in order
👁 Reveal Answer
Correct Answer: B
Explanation: A dead-letter queue (DLQ) receives messages that could not be processed successfully. The source queue’s redrive policy sets maxReceiveCount; once a message has been received that many times without being deleted, SQS moves it to the DLQ, where the team can inspect it and redrive it later. A only slows the retries down. C throws away good messages too. D changes ordering, and a failing message in a FIFO queue can even block the messages behind it in the same group.
Reference: Using dead-letter queues in Amazon SQS
Question 36: Recovery in Under an Hour, at Low Cost
A company’s order system has an RTO of about one hour and an RPO of a few minutes. Running a full copy in a second Region all the time is too expensive. Which disaster recovery strategy fits best?
A) Backup and restore from daily snapshots
B) Multi-site active/active
C) Doing nothing until a disaster happens, then building the environment from scratch
D) Pilot light: keep data continuously replicated to the recovery Region with core infrastructure ready, and start and scale the application servers only during a disaster
👁 Reveal Answer
Correct Answer: D
Explanation: AWS describes four disaster recovery strategies with rising cost and falling recovery time: backup and restore, pilot light, warm standby, and multi-site active/active. Pilot light keeps data live and core pieces provisioned in the recovery Region while most compute is switched off, which supports an RPO of minutes and an RTO in the tens of minutes at low cost. A cannot meet an RPO of minutes with daily snapshots. B meets the target but runs a full second environment, which the company ruled out. C cannot meet either objective.
Reference: Disaster recovery options in the cloud (AWS whitepaper)
Question 37: Switching Users to the Backup Region Automatically
A web application runs in us-east-1 with a standby copy in us-west-2. When the primary site becomes unhealthy, users should be sent to the standby automatically, and back again when the primary recovers. What should the architect configure?
A) Amazon Route 53 failover routing with a health check on the primary endpoint
B) Route 53 simple routing with both IP addresses in one record
C) A single Application Load Balancer spanning both Regions
D) An email alert asking users to change the URL
👁 Reveal Answer
Correct Answer: A
Explanation: Route 53 failover routing supports active-passive configurations. A primary record is associated with a health check, and when the health check fails, Route 53 answers DNS queries with the secondary record, sending users to the standby. When the primary is healthy again, traffic returns to it. B returns both addresses regardless of health. C is not possible, because a load balancer works within a single Region. D is not automatic.
Reference: Active-active and active-passive failover (Amazon Route 53)
Question 38: Copies of Objects in Another Region
A media company must keep a copy of every new object uploaded to an S3 bucket in us-east-1 in a bucket in eu-west-1, automatically and within minutes. What is required?
A) S3 Transfer Acceleration on the source bucket
B) A lifecycle rule that moves objects to S3 Glacier
C) S3 Cross-Region Replication, with versioning enabled on both the source and destination buckets and an IAM role that S3 can use to replicate
D) A CloudFront distribution in front of the bucket
👁 Reveal Answer
Correct Answer: C
Explanation: S3 Replication copies objects automatically to a bucket in another Region (Cross-Region Replication). It requires versioning on both buckets and an IAM role that grants S3 permission to replicate. Live replication applies to new objects after the rule is created; existing objects can be copied with S3 Batch Replication. A speeds up uploads over long distances. B changes storage class in the same bucket. D caches content for viewers but does not create a durable copy.
Reference: Requirements and considerations for replication (Amazon S3)
Question 39: One Place to Manage All Backups
A company backs up EBS volumes with scripts, RDS databases with their own settings, and DynamoDB tables by hand. Auditors want consistent schedules, retention periods, and copies in a second Region, managed from one place. Which service fits?
A) AWS Backup, using backup plans that assign resources by tags and copy recovery points to another Region
B) Amazon S3 Versioning
C) AWS CloudTrail
D) Amazon EC2 Auto Scaling
👁 Reveal Answer
Correct Answer: A
Explanation: AWS Backup centralizes and automates backups across AWS services such as Amazon EBS, Amazon RDS, Amazon DynamoDB, and Amazon EFS. Backup plans define schedules, retention rules, and cross-Region or cross-account copies, and resources can be assigned by tags so new resources are protected automatically. B keeps older versions of S3 objects only. C records API activity. D manages compute capacity.
Reference: Backup plans (AWS Backup)
Question 40: A Relational Database That Survives a Region Outage
A financial application uses Amazon Aurora MySQL in one Region. The business now requires that, if the whole Region becomes unavailable, the database can be running in another Region with only seconds of data loss and in about a minute. What should the architect use?
A) Daily snapshots copied to another Region
B) A Multi-AZ deployment in the same Region
C) A read replica in the same Availability Zone
D) Amazon Aurora Global Database with a secondary cluster in the recovery Region
👁 Reveal Answer
Correct Answer: D
Explanation: Aurora Global Database replicates data from a primary Region to secondary Regions using dedicated infrastructure, with typical replication lag under a second. In a Regional outage, you can fail over by promoting a secondary cluster, which gives a recovery point of seconds and a recovery time of minutes or less. A could lose up to a day of data and takes much longer to restore. B and C protect only within one Region.
Reference: Switchover or failover in Amazon Aurora Global Database
Study Tips for Resilient Architectures
- Multi-AZ is for availability, read replicas are for scale. Multi-AZ is synchronous with automatic failover. Read replicas are asynchronous.
- Make tiers stateless. Move sessions and files off instances into ElastiCache, DynamoDB, EFS, or S3.
- Decouple with queues. SQS absorbs spikes, and dead-letter queues isolate messages that keep failing.
- Know the four DR strategies. Backup and restore, pilot light, warm standby, and multi-site, in order of rising cost and falling RTO and RPO.
- Cross-Region building blocks. Route 53 failover routing, S3 Cross-Region Replication, AWS Backup copies, and Aurora Global Database.
Keep Practicing for SAA-C03
Go back to SAA-C03 Questions 21-30 for data protection, then continue to Questions 41-50 on high-performing architectures. The official SAA-C03 exam guide lists every task in scope.