Exams / SOA-C03 / Study guide / Scaling RDS, Aurora and DynamoDB

Scaling RDS, Aurora and DynamoDB: SOA-C03 study notes and practice questions

What to know about Scaling RDS, Aurora and DynamoDB for the SOA-C03 exam, from the Reliability and Business Continuity domain (22% of the exam guide), then 1 free question to test it. Part of the SOA-C03 study guide.

Last verified October 2026 against current AWS documentation ·How we write our questions

What to know for the exam

Database scaling questions describe a growing load and ask for the least disruptive way to handle it.

Amazon RDS

Vertical scaling: change the instance class (brief outage; Multi-AZ shortens it).
Storage autoscaling: grows storage when free space runs low, up to a maximum you set.
Read replicas: up to 15 for MySQL, MariaDB and PostgreSQL; the application must send reads to the replica endpoints.
RDS Proxy: pools connections for spiky or serverless clients.

Amazon Aurora

Aurora Replicas share the cluster volume; the reader endpoint spreads connections across them.
Aurora Auto Scaling adds and removes replicas with a target tracking policy (for example, average CPU or connections on readers).
Aurora Serverless v2 scales capacity in fine steps with load, and can scale down to zero when idle.
Custom endpoints send specific workloads (such as reporting) to chosen instances.
Aurora Global Database for reads in other Regions.

Amazon DynamoDB

Mode	Scaling
On-demand	Handles traffic changes automatically; pay per request
Provisioned with auto scaling	Target tracking on consumed capacity; reacts within minutes

Throttling (ProvisionedThroughputExceededException) can come from a hot partition even when total capacity is fine. Spread keys more evenly.
DAX caches reads in microseconds.
Global secondary indexes have their own capacity; an under-provisioned GSI can throttle writes to the table.

Common traps

Adding replicas but not changing the app to use them (RDS) or the reader endpoint (Aurora).
Provisioned auto scaling for sudden spikes. It lags; on-demand absorbs spikes.
Ignoring GSI capacity when a table throttles.

Try the 1 practice question

Practice Scaling RDS, Aurora and DynamoDB Take the free practice exam

Question 1

Harder · choose 2 ·Link to this question

A logistics broker runs an Amazon Aurora MySQL cluster with one writer and one Aurora Replica.

Freight-rate reports run from a reporting tool whose demand varies from a few users overnight to hundreds near month-end, and the replica's CPU then reaches 95 percent.

The writer must not be affected by the reports. The broker wants read capacity to follow demand automatically.

Which TWO actions should a CloudOps engineer take? (Choose two.)

Choose two answers.

A. Change the writer to a larger instance class each month-end and change it back afterward with a scheduled Lambda function.
B. Turn on storage autoscaling for the cluster so that Aurora adds I/O capacity to the replica when report demand rises.
C. Point the reporting tool at the cluster endpoint so that Aurora balances its queries between the writer and the replica.
D. Add an Aurora Auto Scaling policy that adds and removes Aurora Replicas to keep average replica CPU utilization near a target.
E. Point the reporting tool at the cluster's reader endpoint so that its connections are spread across all available Aurora Replicas.

Show the answer and explanation

Answers: D, E

A. Wrong: Scaling the writer does not move reports off it and briefly interrupts writes during each change. It also adds custom automation to maintain. Learn more: DB instance classes (opens in a new tab)
B. Wrong: Aurora storage grows automatically and has no storage autoscaling setting to turn on. Storage size is not the bottleneck here, replica CPU is. Learn more: Amazon Aurora storage (opens in a new tab)
C. Wrong: The cluster endpoint always connects to the current writer instance. Reports would run on the writer, which the broker wants to protect. Learn more: Amazon Aurora endpoint connections (opens in a new tab)
D. Right: Correct. Aurora Auto Scaling uses target tracking to change the number of Aurora Replicas. Replicas are added near month-end and removed when demand drops. Learn more: Amazon Aurora Auto Scaling with Aurora Replicas (opens in a new tab)
E. Right: Correct. The reader endpoint distributes connections among the replicas, including ones Aurora Auto Scaling adds. The tool never needs to know individual instance names. Learn more: Amazon Aurora endpoint connections (opens in a new tab)

Explanation

Aurora scales reads horizontally. Aurora Auto Scaling adds and removes replicas with a target tracking policy, and the reader endpoint spreads connections across whatever replicas exist. The cluster endpoint always points at the writer, so reports must use the reader endpoint to stay off it.

Study

Want more on Scaling RDS, Aurora and DynamoDB? The 5 full-length practice sets have 10 more Scaling RDS, Aurora and DynamoDB questions, 325 in all, none of them in the free exam. One payment of $14.99, no subscription. See the practice sets

Study Scaling RDS, Aurora and DynamoDB

Working with DB instance read replicas