Databases
Creating, sizing, and deleting a database
Creating one
A database needs a name, a region and a PostgreSQL version.
The name is the PostgreSQL database name, so it follows PostgreSQL's rules: lowercase letters, numbers and underscores, beginning with a letter. It is also part of every connection URL, so renaming it would break every application connected to it.
The region decides which cluster the database runs in. Pick the one nearest the application that will talk to it, because that distance is added to every query. The region cannot be changed afterwards.
The version is PostgreSQL 15, 16, 17 or 18, defaulting to the newest. Choose an older one only if something you depend on requires it, since the newest gets the longest support life.
What you get
Every database starts on the same allocation and is grown in place when it needs more, so there is no instance type to choose and no capacity planning to do up front.
The database page reports what it actually has, read from the cluster rather than from the record: CPU and memory ranges, storage size, and how many instances are running.
Status
A database reports what it is doing rather than only whether it exists.
| Status | Meaning |
|---|---|
| Provisioning | The cluster is being created. The database is not connectable yet. |
| Active | Ready to use. |
| Error | Provisioning or operation failed. The reported phase says more. |
Usage figures on the database page, such as storage used, query latency and active connections, are measured on the cluster. A measurement that could not be taken is shown as not reported rather than as zero, because a confident zero would be a lie.
Deleting one
Deleting asks you to type the database name to confirm, because the data goes with it.
If the database feeds an analytical mirror, the mirror goes too. See Zero-ETL mirror.