Deploying Models to Azure Analysis Services

Last updated: November 15, 2023

This guide details the various methods and best practices for deploying your Azure Analysis Services models to your Azure environment.

Deployment Methods

There are several ways to deploy your Analysis Services models. The most common methods include:

1. Using Visual Studio with SQL Server Data Tools (SSDT)

Visual Studio with SSDT provides a robust development experience and includes direct deployment capabilities.

1. Open your Analysis Services project in Visual Studio.
2. In the Solution Explorer, right-click on your project.
3. Select Deploy.
4. The Analysis Services Deployment Wizard will appear. Follow the prompts to select your server and database name.

This method is ideal for interactive development and deployment cycles.

2. Using Tabular Editor

Tabular Editor is a popular third-party tool that offers advanced features for managing and deploying Analysis Services models.

• Connect to your Azure Analysis Services instance using Tabular Editor.
• Open your model or a model file (like a TOM object model).
• Make your desired changes and then use the Deploy functionality within Tabular Editor.

Tabular Editor is excellent for scripting deployments and managing complex models.

3. Using Azure DevOps and CI/CD Pipelines

For automated deployments, integrating with Azure DevOps or other CI/CD tools is highly recommended.

• Build Process: Use tasks to build your Analysis Services project (e.g., MSBuild for Visual Studio projects).
• Deployment Task: Utilize specialized tasks for Analysis Services deployment. These might include:
  • Azure CLI tasks with scripts to deploy using TOM (Tabular Object Model).
  • Third-party marketplace extensions for Analysis Services deployment.
  • PowerShell scripts leveraging the Analysis Services Management Objects (ASMO).

This approach ensures consistency, repeatability, and reduces manual errors.

# Example PowerShell snippet for deployment using ASMO
$server = New-Object Microsoft.AnalysisServices.Tabular.Server
$server.Connect("your_azure_analysis_services_server_name.asazure.windows.net")

$database = $server.Databases.GetByName("YourDatabaseName")
$database.Deploy("path/to/your/model.bim", $true) # $true for overwrite
$server.Disconnect()
            

Deployment Considerations

Server Names and Connection Strings

When deploying, you'll need the fully qualified server name. This typically looks like:

Ensure your firewall rules allow access from your deployment agent or development machine.

Database Existence

You can choose to deploy to an existing database or create a new one during deployment. Be mindful of existing data and schema changes.

Overwrite vs. Merge

Most deployment tools offer options to overwrite an existing database or merge changes. Overwriting is simpler but purges existing data. Merging can be more complex but preserves data if done correctly.

Permissions

The account or service principal used for deployment must have the necessary permissions on the Azure Analysis Services server (e.g., Server Administrator role).

Post-Deployment Steps

• Refresh Data: After deployment, you'll typically need to refresh the data within the model. This can be done manually, scheduled via Azure Data Factory, or through custom scripts.
• Testing: Thoroughly test the deployed model by querying it and verifying data integrity and performance.
• Monitoring: Set up monitoring for your Azure Analysis Services instance to track performance, query times, and resource utilization.

By following these guidelines, you can ensure a smooth and efficient deployment process for your Azure Analysis Services models.