SQL support​

Connecting​

EdgeDB supports running read-only SQL queries via the Postgres protocol to enable connecting EdgeDB to existing BI and analytics solutions. Any Postgres-compatible client can connect to your EdgeDB database by using the same port that is used for the EdgeDB protocol and the branch name, username, and password you already use for your database.

Here’s how you might connect to a local instance on port 10701 (determined by running edgedb instance list) on a branch main using the psql CLI:

$ 

psql -h localhost -p 10701 -U edgedb -d main

You’ll then be prompted for a password. If you don’t have it, you can run edgedb instance credentials --insecure-dsn and grab it out of the DSN the command returns. (It’s the string between the second colon and the “at” symbol: edgedb://edgedb:PASSWORD_IS_HERE@<host>:<port>/<branch>)

db> 
... 
... 

alter role edgedb {
  set password := 'my-password'
};

OK: ALTER ROLE

create superuser role sql {
  set password := 'your-password'
};

$ 

psql -h localhost -p 10701 -U sql -d main

Querying​

Object types in your EdgeDB schema are exposed as regular SQL tables containing all the data you store in your EdgeDB database.

If you have a database with the following schema:

module default {
    type Person {
        name: str;
    };

    type Movie extending common::Content {
        release_year: int32;
        director: Person;
        star: Person {
            role: str;
        };
        multi actors: Person {
            role: str;
        };
        multi labels: str;
    };
}
module common {
    type Content {
        title: str;
    };
}

you can access your data after connecting using the following SQL queries:

SELECT id, name FROM "Person";
SELECT id, title, release_year, director_id, star_id FROM "Movie";

Because the link star has link properties, it has its own table. source is the id of the Movie. target is the id of the Person.

SELECT source, target, role FROM "Movie.star";

Links are in separate tables.

SELECT source, target, role FROM "Movie.actors";

Multi properties are in separate tables. source is the id of the Movie. target is the value of the property.

SELECT source, target FROM "Movie.labels";

When types are extended, parent object types’ tables will by default contain all objects of both the type and any types extended by it. The query below will return all common::Content objects as well as all Movie objects.

SELECT id, title FROM common."Content";

To omit objects of extended types, use ONLY. This query will return common::Content objects but not Movie objects.

SELECT id, title FROM ONLY common."Content";

The SQL connector supports read-only statements and will throw errors if the client attempts INSERT, UPDATE, DELETE, or any DDL command. It supports all SQL expressions supported by Postgres.

SELECT id, 'Title is: ' || tittle
FROM "Movie" m
JOIN "Person" d ON m.director_id = d.id
WHERE EXISTS (
    SELECT 1
    FROM "Movie.actors" act
    WHERE act.source = m.id
);

EdgeDB accomplishes this by emulating the information_schema and pg_catalog views to mimic the catalogs provided by Postgres 13.

Tested SQL tools​

• pg_dump

• Metabase

• Cluvio

• Tableau

• DataGrip

• Airbyte [1]

• Fivetran [1]

• Hevo [1]

• Stitch [1]

• dbt [2]

[1] At the moment, EdgeDB does not support “Log replication” (i.e., using the Postgres replication mechanism). Supported replication methods include XMIN Replication, incremental updates using “a user-defined monotonically increasing id,” and full table updates.

[2] dbt models are built and stored in the database as either tables or views. Because the EdgeDB SQL connector does not allow writing or even creating schemas, view, or tables, any attempt to materialize dbt models will result in errors. If you want to build the models, we suggest first transferring your data to a true Postgres instance via pg_dump or Airbyte. Tests and previews can still be run directy against the EdgeDB instance.