Skip to main content
This guide walks through creating an Amazon RDS database with Multi-AZ high availability and a read replica for read scalability. It covers the console workflow, important configuration choices, technical differences between Multi-AZ and read replicas, and cleanup recommendations. Resources:

1 — Open the RDS console and start a new database

In the AWS Console, search for RDS (Relational Database Service) and open the RDS / Amazon Aurora service. The console lists Amazon Aurora alongside the other RDS engine options.
A screenshot of the AWS Aurora/RDS web console showing an empty "Databases (0)" page with a highlighted "Create database" button and a dropdown offering "Express configuration", "Full configuration", and "Restore from S3." The top shows the account/region header and a search/filter bar.
Click Create database. For full control over options required for Multi-AZ and read replicas, choose the Full configuration path rather than Express/Easy Create.

2 — Choose the engine and creation method

Select the database engine that matches your workload—this demo uses PostgreSQL. Choosing PostgreSQL tailors the subsequent options in the wizard.
A screenshot of the AWS RDS "Create database" console showing engine options, with "Aurora (PostgreSQL Compatible)" selected. The page also displays choices like MySQL, PostgreSQL, MariaDB and others.
Under Database creation method select Full configuration.
Screenshot of the AWS RDS "Create database" page showing the "Choose a database creation method" panel with "Full configuration" selected and template options like "Production" and "Dev/Test."
For Templates, choose Dev/Test for a non-production demo (or Production if you require Multi-AZ). Note: the Free Tier template may not support Multi-AZ deployments.

3 — Enable Multi-AZ for high availability

Under Availability and durability select the Multi-AZ DB instance deployment option. This creates a primary instance and a synchronous standby in a different Availability Zone to provide automatic failover and improved durability.
A screenshot of the AWS RDS "Create database" console showing the "Availability and durability" section. It displays deployment options like Multi-AZ DB cluster, Multi-AZ DB instance (selected), and Single-AZ DB instance.
Key points about Multi-AZ:
  • Writes are synchronously replicated to a standby instance in another AZ for data durability.
  • The standby is not a readable endpoint; it exists only for failover/HA.
  • RDS automatically fails over to the standby if the primary becomes unhealthy.
Keep most other settings at their defaults for the demo. Use a current engine version (the demo used PostgreSQL 18.x).
A screenshot of the AWS RDS "Create database" console showing the engine version set to "PostgreSQL 18.3-R2" and options like "Enable RDS Extended Support." The DB instance identifier field is shown with the name "database-1."

4 — Credentials, networking, and public access

For this demo we used a self-managed username and password to keep the steps straightforward. In production environments, integrate with AWS Secrets Manager for managed, rotated credentials. Decide whether the DB should be publicly accessible:
  • For demos, enabling public access gives an easy endpoint for testing.
  • For production, avoid public exposure—use private subnets, VPC security groups, and IAM/Secrets Manager integration.
Enabling public access exposes the DB endpoint to the Internet and can be a security risk. For production, prefer private subnets, VPC security groups, and Secrets Manager for credentials.

5 — Create and wait for the instance

Create the database and wait for provisioning. The console will show the instance transitioning from Creating to Available.
A screenshot of the AWS RDS (Aurora and RDS) console with a blue banner saying "Creating database database-1" and a "View connection details" button. The Databases list shows an entry named database-1 with status "Creating" and engine PostgreSQL.
After creation, the console will show the DB instance with Multi-AZ = Yes. The standby instance is managed by RDS and does not appear as a separate endpoint.

6 — Read replicas: when and how to add them

Read replicas are different from Multi-AZ: Multi-AZ provides availability (standby, synchronous replication), while read replicas provide read scaling (usually asynchronous replication and readable endpoints).
Multi-AZ = high availability (synchronous replication, standby not readable). Read replicas = read scaling (typically asynchronous replication, readable endpoints).
Before creating a read replica, confirm the source DB is in the Available state (or at least backing up); otherwise the Create read replica action may be disabled. To create a read replica:
  1. Select the source database in the console.
  2. Choose Actions → Create read replica.
  3. Set a DB instance identifier (for example database-1-readreplica).
  4. Keep most settings at defaults; the demo used a Single-AZ replica to reduce cost and complexity.
A screenshot of the AWS RDS "Create read replica" settings page showing the replica source set to "database-1" and the DB instance identifier field filled with "database-1-readreplica." The browser top bar and AWS navigation are visible, including the CloudWithDjango-Org account name.
Choose the availability option for the replica (Single-AZ in this demo). Read replicas are intended to offload read-heavy traffic—reporting, analytics, and read-only queries—away from the primary.
Screenshot of the AWS RDS "Create read replica" page showing Availability deployment options. It displays choices for Multi-AZ DB cluster, Multi-AZ DB instance, and a highlighted Single-AZ DB instance deployment.
Important technical distinctions:
  • Standard RDS read replicas (MySQL, MariaDB, PostgreSQL) generally use asynchronous replication — expect replication lag.
  • Aurora replicas use a different, storage-based replication design that offers faster scaling and lower lag.
  • Read replicas are readable and can serve application read traffic. They are not automatically promoted on primary failure unless you manually promote them or implement failover automation.
  • Use Multi-AZ for automatic failover/availability. Use read replicas to scale read capacity horizontally.

7 — Wait for the replica and verify

Create the read replica and wait until its status is Available. The console will list the primary as Primary and the replica(s) as Replica(s).
A screenshot of the AWS RDS/Aurora console with a green banner reading "Successfully created replica databasereadreplica in US East (Ohio)". The databases list shows "database-1" as the Primary and "databasereadreplica" as a Replica, both marked Available and using PostgreSQL.
Route read traffic to read replicas by using a separate connection string or by integrating a connection manager/load-balancer within your application logic. Monitor replication lag and performance to ensure read replicas meet your SLA.

Quick comparison: Multi-AZ vs Read Replica

Best practices and cleanup

  • Use Secrets Manager for credentials and rotate them regularly.
  • Keep DB instances in private subnets and control access with VPC security groups.
  • Monitor replication lag, CPU, memory, and disk I/O on primary and replicas.
  • When tearing down test resources: delete read replicas first, then the primary. Decide whether to take a final snapshot—skip for throwaway demos, but keep snapshots for production or important data.
Remember to delete demo resources to avoid ongoing charges. Thank you — this guide covered creating an RDS Multi-AZ DB instance for high availability and adding a read replica for read scalability.

Watch Video