Databases

Alongside the catalog, the platform runs shared Postgres instances and Cassandra clusters. Your organisation does not get an instance of its own: it gets a schema on a Postgres instance, or a keyspace on a Cassandra cluster, that only it can use.

• A Postgres schema
• A Cassandra keyspace
• Querying from a notebook
• Instances and clusters

A Postgres schema

Search the command palette for Postgres Schemas and choose New schema. Choose a name and the instance it lives on. The platform creates the schema for your organisation, and a Postgres role that can reach that schema and nothing else. Owners, admins and members can create one.

A Cassandra keyspace

Search the command palette for Cassandra Keyspaces and choose New keyspace. Choose a name and the cluster it lives on. The keyspace belongs to your organisation alone. Owners, admins and members can create one.

Querying from a notebook

Both are queried through your organisation's SQL warehouse, with the %%trino command in a kernel notebook. You sign in as yourself — there is no database username or password to manage.

Database	Address a table as
Postgres	postgresql.<schema>.<table>
Cassandra	cassandra.<keyspace>.<table>

For example, in a notebook attached to a kernel:

%%trino -o events
SELECT * FROM cassandra.analytics.events LIMIT 100

A SQL warehouse must be running for %%trino to work. See SQL from a Python notebook for its options.

If a schema or keyspace was created while the access service could not be reached, the page says so and names a reference. It exists, but queries through the warehouse are refused until an administrator re-applies access.

Instances and clusters

The Postgres and Cassandra pages, under Databases in the main menu, list the shared instances and clusters, their state, size and version, with what they cost this month across the platform. Creating, resizing, starting and stopping them is for owners and admins. A Cassandra cluster scales its nodes with load between a minimum and a maximum, and can scale to zero.