Skip to main content

Microsoft Fabric Synapse Data Warehouse setup

info

Below is a guide for use with Synapse Data Warehouse, a new product within Microsoft Fabric. The adapter currently only supports connecting to a warehouse and not a lakehouse endpoint. You can access data in your lakehouse via the warehouse if you are in the same workspace.

To learn how to set up dbt with Azure Synapse Dedicated Pools, refer to Microsoft Azure Synapse DWH setup.

  • Maintained by: Microsoft
  • Authors: Microsoft
  • GitHub repo: Microsoft/dbt-fabric
  • PyPI package: dbt-fabric
  • Slack channel:
  • Supported dbt Core version: 1.4.0 and newer
  • dbt Cloud support: Supported
  • Minimum data platform version:

Installing dbt-fabric

Use pip to install the adapter. Before 1.8, installing the adapter would automatically install dbt-core and any additional dependencies. Beginning in 1.8, installing an adapter does not automatically install dbt-core. This is because adapters and dbt Core versions have been decoupled from each other so we no longer want to overwrite existing dbt-core installations. Use the following command for installation:

Configuring dbt-fabric

For Microsoft Fabric-specific configuration, please refer to Microsoft Fabric configs.

Prerequisites

On Debian/Ubuntu make sure you have the ODBC header files before installing

sudo apt install unixodbc-dev

Download and install the Microsoft ODBC Driver 18 for SQL Server. If you already have ODBC Driver 17 installed, then that one will work as well.

Supported configurations

  • The adapter is tested with Microsoft Fabric Synapse Data Warehouses (also referred to as Warehouses).
  • We test all combinations with Microsoft ODBC Driver 17 and Microsoft ODBC Driver 18.
  • The collations we run our tests on are Latin1_General_100_BIN2_UTF8.

The adapter support is not limited to the matrix of the above configurations. If you notice an issue with any other configuration, let us know by opening an issue on GitHub.

Unsupported configurations

SQL analytics endpoints are read-only and so are not appropriate for Transformation workloads, use a Warehouse instead.

Authentication methods & profile configuration

Common configuration

For all the authentication methods, refer to the following configuration options that can be set in your profiles.yml file. A complete reference of all options can be found at the end of this page.

Configuration optionDescriptionTypeExample
driverThe ODBC driver to useRequiredODBC Driver 18 for SQL Server
serverThe server hostnameRequiredlocalhost
portThe server portRequired1433
databaseThe database nameRequiredNot applicable
schemaThe schema nameRequireddbo
retriesThe number of automatic times to retry a query before failing. Defaults to 1. Queries with syntax errors will not be retried. This setting can be used to overcome intermittent network issues.OptionalNot applicable
login_timeoutThe number of seconds used to establish a connection before failing. Defaults to 0, which means that the timeout is disabled or uses the default system settings.OptionalNot applicable
query_timeoutThe number of seconds used to wait for a query before failing. Defaults to 0, which means that the timeout is disabled or uses the default system settings.OptionalNot applicable
schema_authorizationOptionally set this to the principal who should own the schemas created by dbt. Read more about schema authorization.OptionalNot applicable
encryptWhether to encrypt the connection to the server. Defaults to true. Read more about connection encryption.OptionalNot applicable
trust_certWhether to trust the server certificate. Defaults to false. Read more about connection encryption.OptionalNot applicable

Connection encryption

Microsoft made several changes in the release of ODBC Driver 18 that affects how connection encryption is configured. To accommodate these changes, starting in dbt-sqlserver 1.2.0 or newer the default values of encrypt and trust_cert have changed. Both of these settings will now always be included in the connection string to the server, regardless if you've left them out of your profile configuration or not.

  • The default value of encrypt is true, meaning that connections are encrypted by default.
  • The default value of trust_cert is false, meaning that the server certificate will be validated. By setting this to true, a self-signed certificate will be accepted.

More details about how these values affect your connection and how they are used differently in versions of the ODBC driver can be found in the Microsoft documentation.

Standard SQL Server authentication

SQL Server and windows authentication are not supported by Microsoft Fabric Synapse Data Warehouse.

Microsoft Entra ID authentication

Microsoft Entra ID (formerly Azure AD) authentication is a default authentication mechanism in Microsoft Fabric Synapse Data Warehouse.

The following additional methods are available to authenticate to Azure SQL products:

  • Microsoft Entra ID username and password
  • Service principal
  • Environment-based authentication
  • Azure CLI authentication
  • VS Code authentication (available through the automatic option below)
  • Azure PowerShell module authentication (available through the automatic option below)
  • Automatic authentication

The automatic authentication setting is in most cases the easiest choice and works for all of the above.

First, install the Azure CLI, then, log in:

az login

profiles.yml
your_profile_name:
target: dev
outputs:
dev:
type: fabric
driver: 'ODBC Driver 18 for SQL Server' # (The ODBC Driver installed on your system)
server: hostname or IP of your server
port: 1433
database: exampledb
schema: schema_name
authentication: CLI

Additional options for Microsoft Entra ID on Windows

On Windows systems, the following additional authentication methods are also available for Azure SQL:

  • Microsoft Entra ID interactive
  • Microsoft Entra ID integrated
  • Visual Studio authentication (available through the automatic option above)

This setting can optionally show Multi-Factor Authentication prompts.

profiles.yml
your_profile_name:
target: dev
outputs:
dev:
type: fabric
driver: 'ODBC Driver 18 for SQL Server' # (The ODBC Driver installed on your system)
server: hostname or IP of your server
port: 1433
database: exampledb
schema: schema_name
authentication: ActiveDirectoryInteractive
user: bill.gates@microsoft.com

Automatic Microsoft Entra ID principal provisioning for grants

Please note that automatic Microsoft Entra ID principal provisioning is not supported by Microsoft Fabric Synapse Data Warehouse at this time. Even though in dbt 1.2 or newer you can use the grants config block to automatically grant/revoke permissions on your models to users or groups, the data warehouse does not support this feature at this time.

You need to add the service principal or Microsoft Entra identity to a Fabric Workspace as an admin

Schema authorization

You can optionally set the principal who should own all schemas created by dbt. This is then used in the CREATE SCHEMA statement like so:

CREATE SCHEMA [schema_name] AUTHORIZATION [schema_authorization]

A common use case is to use this when you are authenticating with a principal who has permissions based on a group, such as a Microsoft Entra ID group. When that principal creates a schema, the server will first try to create an individual login for this principal and then link the schema to that principal. If you would be using Microsoft Entra ID in this case, then this would fail since Azure SQL can't create logins for individuals part of an AD group automatically.

Reference of all connection options

Configuration optionDescriptionRequiredDefault value
driverThe ODBC driver to use.
hostThe hostname of the database server.
portThe port of the database server.1433
databaseThe name of the database to connect to.
schemaThe schema to use.
authenticationThe authentication method to use. This is not required for Windows authentication.'sql'
UIDUsername used to authenticate. This can be left out depending on the authentication method.
PWDPassword used to authenticate. This can be left out depending on the authentication method.
tenant_idThe tenant ID of the Microsoft Entra ID instance. This is only used when connecting to Azure SQL with a service principal.
client_idThe client ID of the Microsoft Entra service principal. This is only used when connecting to Azure SQL with a Microsoft Entra service principal.
client_secretThe client secret of the Microsoft Entra service principal. This is only used when connecting to Azure SQL with a Microsoft Entra service principal.
encryptSet this to false to disable the use of encryption. See above.true
trust_certSet this to true to trust the server certificate. See above.false
retriesThe number of times to retry a failed connection.1
schema_authorizationOptionally set this to the principal who should own the schemas created by dbt. Details above.
login_timeoutThe amount of seconds to wait until a response from the server is received when establishing a connection. 0 means that the timeout is disabled.0
query_timeoutThe amount of seconds to wait until a response from the server is received when executing a query. 0 means that the timeout is disabled.0

Valid values for authentication:

  • ActiveDirectoryPassword: Active Directory authentication using username and password
  • ActiveDirectoryInteractive: Active Directory authentication using a username and MFA prompts
  • ActiveDirectoryIntegrated: Active Directory authentication using the current user's credentials
  • ServicePrincipal: Microsoft Entra ID authentication using a service principal
  • CLI: Microsoft Entra ID authentication using the account you're logged in within the Azure CLI
  • environment: Microsoft Entra ID authentication using environment variables as documented here
  • auto: Microsoft Entra ID authentication trying the previous authentication methods until it finds one that works
0