Edmund AI Documentation
/DOCUMENTATION

Specific Database Considerations

Microsoft SQL Server

For Microsoft SQL Server it is necessary to only have one user, with all the tables that Edmund has to query being inside one schema. This schema should also be the default schema for the user that Edmund will be using to connect to the database. Here you have two options:

  • Use the already predefined default schema dbo for your tables

  • Create your own schema and assign as default to the user. How to set this up can be found in the official Microsoft documentation.

Oracle Database

By default, synonyms are turned on in Oracle databases. From our experience, we prefer to use tables to views.

All tables or views Edmund has access to must be present inside one schema. If you want to use a schema different from the user, you have these options:

  • Set up a logon trigger for the SQL database like described here.

  • Use synonyms to link tables to the user’s schema.

PostgreSQL

For PostgreSQL it is necessary to have all the tables or views that Edmund queries present inside one schema. This schema should also be the default schema for the user that Edmund will be using to connect to the database, meaning it should be the first entry in that user’s search_path. Here you have two options:

  • Use the already predefined default schema public for your tables

  • Create your own schema and set it as the default for the user by adjusting its search_path (for example with ALTER ROLE … SET search_path). How to set this up can be found in the official PostgreSQL documentation.

In both cases, make sure the connecting user has been granted USAGE on the schema and SELECT on the relevant tables or views.

Snowflake

Edmund connects to Snowflake with your account identifier, a user, and the database, schema and warehouse it should use. A role is optional. All tables or views that Edmund queries must be in that one schema.

  • Authenticate with key-pair authentication (recommended) or a password.

  • Grant the role USAGE on the warehouse, database and schema, and SELECT on the relevant tables or views.

  • Edmund’s queries run on the warehouse you specify and consume its credits. We recommend a small dedicated warehouse with auto-suspend.

SAP HANA

Edmund connects to the SQL port of your HANA system. In a multi-tenant system, specify the tenant database name. All tables or views that Edmund queries must be in one schema, which you set on the data source. The connection is encrypted by default.

  • Create a dedicated user and grant it SELECT on that schema, for example GRANT SELECT ON SCHEMA <schema> TO <user>.

  • SAP HANA is enabled per organization. Ask Edmund to turn it on before you set up the data source.

InfluxDB

Edmund supports InfluxDB 2.x and connects with the instance URL, your organization and an API token. InfluxDB has no schemas, so Edmund reads the structure (buckets, measurements, tags and fields) of everything the token can access.

  • Use a read-only API token scoped to the buckets Edmund should see.