Platform
Locations and regions
You choose a location; the platform picks the AWS region in it that offers what's needed, with fallbacks.
Locations
A location is a place you choose for a project or a database or bucket, such as a city. Each location has an ordered list of AWS regions: the first is its primary region and the others are fallbacks. The list of locations, and which one is the default, is maintained by the platform team; Cloud shows the current ones wherever you pick a location.
How a region is chosen
When something needs a region, the platform asks for the first region in the location's list that is active and offers the service:
| What | Service asked for | When |
|---|---|---|
| A Next.js deployment's server function | lambda | Each time a deployment is published, so new deployments follow registry changes. |
| A database | dynamodb | When it's created. It stays in that region. |
| A bucket | s3 | When it's created. It stays in that region. |
| Chat | bedrock | For every conversation, at the default location. |
So if a location's primary region doesn't offer a service, or is taken out of service, that service runs in the next region on the list.
Repositories are stored in the platform's home region, us-west-1, whatever location you pick. Static files and the edge (CloudFront) are global.
Where things show up
- A project's location is shown on its page and can't be changed after creation.
- A deployment's region is shown on the deployment page.
- Databases and buckets show their location and region in Storage.
- Linked storage passes its region to your code as
SI_DATABASE_<NAME>_REGION/SI_BUCKET_<NAME>_REGION(see Environment variables).
Registry changes reach running services within a minute.