RDS Storage Type Change: Downtime for gp2 → gp3 and gp3 → gp2?

Hello There,

I would like to inform you that changing the storage type from gp2 to gp3 or gp3 to gp2 does not cause downtime. The only downtime that is expected to occur from a change of storage type is from one of these types to magnetic and vice-versa. If you set apply_immediately = true, the storage type change will begin immediately.

[+] : AWS - Amazon RDS storage types - https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/CHAP_Storage.html#Concepts.Storage

To shed some more light on, When you initiate the change to the storage type, your instance will go into the "storage-optimization" status. In this mode, Amazon RDS is optimizing the storage of your DB instance. The DB instance is fully operational. The amount of time this process can take is variable and depends on a variety of factors - the storage optimization process is usually short, but can sometimes take up to and even beyond 24 hours.

Important Note: Please be advised that after Amazon RDS begins to modify your DB instance to change the storage size or type, you can't submit another request to change the storage size, performance, or type for six hours.

[+] : AWS - Viewing Amazon RDS DB instance status - https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/accessing-monitoring.html

Best Practices:

• Perform the change during a low-traffic period to minimize potential performance impact.
• Monitor your instance's performance metrics before and after the change to ensure the desired outcome.
• Test the change in a non-production environment first if possible.
• Ensure you have recent backups before making any changes.
• If you're changing to gp3, consider adjusting the IOPS and throughput settings to optimize performance for your workload.

It's important to note that while the storage type change itself doesn't cause downtime, other simultaneous modifications or specific configurations might. For instance, if you're using a DB instance class with NVMe-based storage (like r6gd), changing storage configurations may require a restart.

Always review the specific documentation for your DB engine and version, as there might be engine-specific considerations. Additionally, while the change process is designed to be non-disruptive, it's always wise to plan for potential unexpected issues and have a rollback plan ready.

Sources
Downtime of Switching Storage Configuration on Aurora PostgreSQL with NVMe-based Reader Instance | AWS re:Post
Identify causes of downtime for an RDS or Aurora instance | AWS re:Post
How to change volumes specs on RTS for better performance? | AWS re:Post