Skip to main content
Create one SQL database resource with PostgreSQL or MySQL. Choose standard Amazon RDS or Aurora. Existing Aurora Serverless databases remain supported. Spring Winter creates the database in your project network. The database is private and encrypted. Available engines, versions, compute classes, and storage options depend on your AWS Region.

Choose a database

Choose Aurora Serverless for workloads that need serverless capacity. Choose Aurora provisioned for fixed instance capacity or reader nodes. Choose standard RDS when you need an RDS instance with storage and availability settings. A larger writer provides more write capacity. A reader provides read capacity when your application sends reads to its endpoint. An RDS Multi-AZ standby provides failover. It does not serve reads.

Increase capacity

Open the database, then select Settings. Review the change before you submit it.
  • Change the writer instance class to increase compute capacity.
  • Add, resize, or remove reader nodes where supported.
  • Increase RDS storage or change supported storage performance settings. Storage cannot shrink.
  • Enable or disable an RDS Multi-AZ standby.
  • Change supported Aurora Serverless capacity limits.
  • Convert Aurora writer compute between serverless and provisioned modes.
Some changes start at once. Other changes can wait for the next maintenance window. AWS can interrupt connections during a change. Your application must reconnect. Spring Winter shows accepted, scheduled, running, completed, or failed progress based on the workflow and observed AWS state. An accepted request does not mean that AWS has completed the change. Aurora compute conversion uses a ready reader and a controlled failover. The former writer can remain as a reader until you remove it. You pay for each active database node. Spring Winter does not enable automatic reader scaling.

Connect to the database

The writer endpoint accepts writes. Aurora also provides a reader endpoint when readers are available. Standard RDS reader endpoints depend on the configured read replica. Your application controls read routing. Use the database Connect section to fetch credentials or open a remote tunnel when available. Standard RDS passwords stay in AWS Secrets Manager in your account. Spring Winter does not store the password in its application database. Password rotation is not automatic. Spring Winter does not set your application’s database URL. Add the connection details to your application using your own secret-management process. Configure your database client to use TLS and verify the server certificate and hostname. Check the settings for your app’s driver and make the appropriate AWS RDS CA certificate available to the client. An encrypted connection can still lack server identity verification.

Metrics and auditing

The Metrics page shows available node, storage, and replication metrics. Metric availability depends on the database family and AWS configuration. Connection auditing depends on supported RDS parameter and option groups. Spring Winter reports when an audit capability is unavailable. Standard RDS MySQL connection auditing is not available in this release. Some PostgreSQL audit changes can require a reboot.

Backups and deletion

Choose the deletion policy before you delete a database. The API keeps a final snapshot by default. A retained snapshot can incur AWS storage charges. For disposable data, pass delete_backup=true only when skipping the final snapshot is part of the approved cleanup. Spring Winter deletes reader nodes before the primary database. For standard RDS, the credential secret is retained with the final snapshot according to the selected recovery policy. If a new database creation reaches a terminal failure, Spring Winter retries temporary failures and then starts cleanup for resources from that creation. Cleanup does not create a final snapshot. The failed resource remains visible until cleanup succeeds. If cleanup fails, retry it from the database’s Delete settings. Automatic cleanup does not apply to older failed databases.

Cost

The database page shows resource cost information when AWS has reported it. The project Cost page shows billed AWS usage for the trailing 30 days. AWS cost data can take about one day to appear. See Costs and estimates for the difference between estimates and billed usage.