databricks-2025

name	databricks-2025
description	Databricks Job activity and 2025 Azure Data Factory connectors

🚨 CRITICAL GUIDELINES

Windows File Path Requirements

MANDATORY: Always Use Backslashes on Windows for File Paths

When using Edit or Write tools on Windows, you MUST use backslashes (\) in file paths, NOT forward slashes (/).

Examples:

• ❌ WRONG: D:/repos/project/file.tsx
• ✅ CORRECT: D:\repos\project\file.tsx

This applies to:

• Edit tool file_path parameter
• Write tool file_path parameter
• All file operations on Windows systems

Documentation Guidelines

NEVER create new documentation files unless explicitly requested by the user.

• Priority: Update existing README.md files rather than creating new documentation
• Repository cleanliness: Keep repository root clean - only README.md unless user requests otherwise
• Style: Documentation should be concise, direct, and professional - avoid AI-generated tone
• User preference: Only create additional .md files when user specifically asks for documentation

---

Azure Data Factory Databricks Integration 2025

Databricks Job Activity (Recommended 2025)

🚨 CRITICAL UPDATE (2025): The Databricks Job activity is now the ONLY recommended method for orchestrating Databricks in ADF. Microsoft strongly recommends migrating from legacy Notebook, Python, and JAR activities.

Why Databricks Job Activity?

Old Pattern (Notebook Activity - ❌ LEGACY):

{
  "name": "RunNotebook",
  "type": "DatabricksNotebook",  // ❌ DEPRECATED - Migrate to DatabricksJob
  "linkedServiceName": { "referenceName": "DatabricksLinkedService" },
  "typeProperties": {
    "notebookPath": "/Users/[e-mail]/MyNotebook",
    "baseParameters": { "param1": "value1" }
  }
}

New Pattern (Databricks Job Activity - ✅ CURRENT 2025):

{
  "name": "RunDatabricksWorkflow",
  "type": "DatabricksJob",  // ✅ CORRECT activity type (NOT DatabricksSparkJob)
  "linkedServiceName": { "referenceName": "DatabricksLinkedService" },
  "typeProperties": {
    "jobId": "123456",  // Reference existing Databricks Workflow Job
    "jobParameters": {  // Pass parameters to the Job
      "param1": "value1",
      "runDate": "@pipeline().parameters.ProcessingDate"
    }
  },
  "policy": {
    "timeout": "0.12:00:00",
    "retry": 2,
    "retryIntervalInSeconds": 30
  }

Benefits of Databricks Job Activity (2025)

1. Serverless Execution by Default:

   • ✅ No cluster specification needed in linked service
   • ✅ Automatically runs on Databricks serverless compute
   • ✅ Faster startup times and lower costs
   • ✅ Managed infrastructure by Databricks

2. Advanced Workflow Features:

   • ✅ Run As - Execute jobs as specific users/service principals
   • ✅ Task Values - Pass data between tasks within workflow
   • ✅ Conditional Execution - If/Else and For Each task types
   • ✅ AI/BI Tasks - Model serving endpoints, Power BI semantic models
   • ✅ Repair Runs - Rerun failed tasks without reprocessing successful ones
   • ✅ Notifications/Alerts - Built-in alerting on job failures
   • ✅ Git Integration - Version control for notebooks and code
   • ✅ DABs Support - Databricks Asset Bundles for deployment
   • ✅ Built-in Lineage - Data lineage tracking across tasks
   • ✅ Queuing and Concurrent Runs - Better resource management

3. Centralized Job Management:

   • Jobs defined once in Databricks workspace
   • Single source of truth for all environments
   • Versioning through Databricks (Git-backed)
   • Consistent across orchestration tools

4. Better Orchestration:

   • Complex task dependencies within Job
   • Multiple heterogeneous tasks (notebook, Python, SQL, Delta Live Tables)
   • Job-level monitoring and logging
   • Parameter passing between tasks

5. Improved Reliability:

   • Retry logic at Job and task level
   • Better error handling and recovery
   • Automatic cluster management

6. Cost Optimization:

   • Serverless compute (pay only for execution)
   • Job clusters (auto-terminating)
   • Optimized cluster sizing per task
   • Spot instance support

Implementation

1. Create Databricks Job

# In Databricks workspace
# Create Job with tasks
{
  "name": "Data Processing Job",
  "tasks": [
    {
      "task_key": "ingest",
      "notebook_task": {
        "notebook_path": "/Notebooks/Ingest",
        "base_parameters": {}
      },
      "job_cluster_key": "small_cluster"
    },
    {
      "task_key": "transform",
      "depends_on": [{ "task_key": "ingest" }],
      "notebook_task": {
        "notebook_path": "/Notebooks/Transform"
      },
      "job_cluster_key": "medium_cluster"
    },
    {
      "task_key": "load",
      "depends_on": [{ "task_key": "transform" }],
      "notebook_task": {
        "notebook_path": "/Notebooks/Load"
      },
      "job_cluster_key": "small_cluster"
    }
  ],
  "job_clusters": [
    {
      "job_cluster_key": "small_cluster",
      "new_cluster": {
        "spark_version": "13.3.x-scala2.12",
        "node_type_id": "Standard_DS3_v2",
        "num_workers": 2
      }
    },
    {
      "job_cluster_key": "medium_cluster",
      "new_cluster": {
        : ,
        : ,
        : 
      }
    }
  ]
}

2. Create ADF Pipeline with Databricks Job Activity (2025)

{
  "name": "PL_Databricks_Serverless_Workflow",
  "properties": {
    "activities": [
      {
        "name": "ExecuteDatabricksWorkflow",
        "type": "DatabricksJob",  // ✅ Correct activity type
        "dependsOn": [],
        "policy": {
          "timeout": "0.12:00:00",
          "retry": 2,
          "retryIntervalInSeconds": 30
        },
        "typeProperties": {
          "jobId": "123456",  // Databricks Job ID from workspace
          "jobParameters": {  // ⚠️ Use jobParameters (not parameters)
             
             
             
             
          
        
         
           
           
        
      
      
         
         
         
          
             
             
          
        
         
           
           
           
             
             
             
             
          
        
      
    
     
       
         
         
      
       
         
         
      
       
         
      
    
  

3. Configure Linked Service (2025 - Serverless)

✅ RECOMMENDED: Serverless Linked Service (No Cluster Configuration)

{
  "name": "DatabricksLinkedService_Serverless",
  "type": "Microsoft.DataFactory/factories/linkedservices",
  "properties": {
    "type": "AzureDatabricks",
    "typeProperties": {
      "domain": "https://adb-123456789.azuredatabricks.net",
      "authentication": "MSI"  // ✅ Managed Identity (recommended 2025)
      // ⚠️ NO existingClusterId or newClusterNodeType needed for serverless!
      // The Databricks Job activity automatically uses serverless compute
    }
  }
}

Alternative: Access Token Authentication

{
  "name": "DatabricksLinkedService_Token",
  "type": "Microsoft.DataFactory/factories/linkedservices",
  "properties": {
    "type": "AzureDatabricks",
    "typeProperties": {
      "domain": "https://adb-123456789.azuredatabricks.net",
      "accessToken": {
        "type": "AzureKeyVaultSecret",
        "store": {
          "referenceName": "AzureKeyVault",
          "type": "LinkedServiceReference"
        },
        "secretName": "databricks-access-token"
      }
    }
  }
}

🚨 CRITICAL: For Databricks Job activity, DO NOT specify cluster properties in the linked service. The job configuration in Databricks workspace controls compute resources.

🆕 2025 New Connectors and Enhancements

ServiceNow V2 Connector (RECOMMENDED - V1 End of Support)

🚨 CRITICAL: ServiceNow V1 connector is at End of Support stage. Migrate to V2 immediately!

Key Features of V2:

• ✅ Native Query Builder - Aligns with ServiceNow's condition builder experience
• ✅ Enhanced Performance - Optimized data extraction
• ✅ Better Error Handling - Improved diagnostics and retry logic
• ✅ OData Support - Modern API integration patterns

Copy Activity Example:

{
  "name": "CopyFromServiceNowV2",
  "type": "Copy",
  "inputs": [
    {
      "referenceName": "ServiceNowV2Source",
      "type": "DatasetReference"
    }
  ],
  "outputs": [
    {
      "referenceName": "AzureSqlSink",
      "type": "DatasetReference"
    }
  ],
  "typeProperties": {
    "source": {
      "type": "ServiceNowV2Source",
      "query": "sysparm_query=active=true^priority=1^sys_created_on>=javascript:gs.dateGenerate('2025-01-01')",
      "httpRequestTimeout": "00:01:40"  // 100 seconds
    
     
       
       
       
         
         
      
    
     
     
       
         
         
      
    
  

Linked Service (OAuth2 - Recommended):

{
  "name": "ServiceNowV2LinkedService",
  "type": "Microsoft.DataFactory/factories/linkedservices",
  "properties": {
    "type": "ServiceNowV2",
    "typeProperties": {
      "endpoint": "https://dev12345.service-now.com",
      "authenticationType": "OAuth2",
      "clientId": "your-oauth-client-id",
      "clientSecret": {
        "type": "AzureKeyVaultSecret",
        "store": {
          "referenceName": "AzureKeyVault",
          "type": "LinkedServiceReference"
        },
        "secretName": "servicenow-client-secret"
      }
       
       
         
         
           
           
        
         
      
       
    
  

Linked Service (Basic Authentication - Legacy):

{
  "name": "ServiceNowV2LinkedService_Basic",
  "type": "Microsoft.DataFactory/factories/linkedservices",
  "properties": {
    "type": "ServiceNowV2",
    "typeProperties": {
      "endpoint": "https://dev12345.service-now.com",
      "authenticationType": "Basic",
      "username": "admin",
      "password": {
        "type": "AzureKeyVaultSecret",
        "store": {
          "referenceName": "AzureKeyVault",
          "type": "LinkedServiceReference"
        },
        "secretName": "servicenow-password"
      }
    }
  

Migration from V1 to V2:

1. Update linked service type from ServiceNow to ServiceNowV2
2. Update source type from ServiceNowSource to ServiceNowV2Source
3. Test queries in ServiceNow UI's condition builder first
4. Adjust timeout settings if needed (V2 may have different performance)

Enhanced PostgreSQL Connector

Improved performance and features:

{
  "name": "PostgreSQLLinkedService",
  "type": "PostgreSql",
  "typeProperties": {
    "connectionString": "host=myserver.postgres.database.azure.com;port=5432;database=mydb;uid=myuser",
    "password": {
      "type": "AzureKeyVaultSecret",
      "store": { "referenceName": "KeyVault" },
      "secretName": "postgres-password"
    },
    // 2025 enhancement
    "enableSsl": true,
    "sslMode": "Require"
  }
}

Microsoft Fabric Warehouse Connector (NEW 2025)

🆕 Native support for Microsoft Fabric Warehouse (Q3 2024+)

Supported Activities:

• ✅ Copy Activity (source and sink)
• ✅ Lookup Activity
• ✅ Get Metadata Activity
• ✅ Script Activity
• ✅ Stored Procedure Activity

Linked Service Configuration:

{
  "name": "FabricWarehouseLinkedService",
  "type": "Microsoft.DataFactory/factories/linkedservices",
  "properties": {
    "type": "Warehouse",  // ✅ NEW dedicated Fabric Warehouse type
    "typeProperties": {
      "endpoint": "myworkspace.datawarehouse.fabric.microsoft.com",
      "warehouse": "MyWarehouse",
      "authenticationType": "ServicePrincipal",  // Recommended
      "servicePrincipalId": "<app-registration-id>",
      "servicePrincipalKey": {
        "type": "AzureKeyVaultSecret",
        "store": {
          "referenceName": "AzureKeyVault",
          "type": "LinkedServiceReference"
        
         
      
       
    
  

Alternative: Managed Identity Authentication (Preferred)

{
  "name": "FabricWarehouseLinkedService_ManagedIdentity",
  "type": "Microsoft.DataFactory/factories/linkedservices",
  "properties": {
    "type": "Warehouse",
    "typeProperties": {
      "endpoint": "myworkspace.datawarehouse.fabric.microsoft.com",
      "warehouse": "MyWarehouse",
      "authenticationType": "SystemAssignedManagedIdentity"
    }
  }
}

Copy Activity Example:

{
  "name": "CopyToFabricWarehouse",
  "type": "Copy",
  "inputs": [
    {
      "referenceName": "AzureSqlSource",
      "type": "DatasetReference"
    }
  ],
  "outputs": [
    {
      "referenceName": "FabricWarehouseSink",
      "type": "DatasetReference"
    }
  ],
  "typeProperties": {
    "source": {
      "type": "AzureSqlSource"
    },
    "sink": {
      "type": "WarehouseSink",
      "writeBehavior":   
       
         
    
       
     
       
         
         
      
       
    
     
       
       
        
              
              
        
      
    
  

Best Practices for Fabric Warehouse:

• ✅ Use managed identity for authentication (no secret rotation)
• ✅ Enable staging for large data loads (> 1GB)
• ✅ Use tableOption: autoCreate for dynamic schema creation
• ✅ Leverage Fabric's lakehouse integration for unified analytics
• ✅ Monitor Fabric capacity units (CU) consumption

Enhanced Snowflake Connector

Improved performance:

{
  "name": "SnowflakeLinkedService",
  "type": "Snowflake",
  "typeProperties": {
    "connectionString": "jdbc:snowflake://myaccount.snowflakecomputing.com",
    "database": "mydb",
    "warehouse": "mywarehouse",
    "authenticationType": "KeyPair",
    "username": "myuser",
    "privateKey": {
      "type": "AzureKeyVaultSecret",
      "store": { "referenceName": "KeyVault" },
      "secretName": "snowflake-private-key"
    },
    "privateKeyPassphrase": {
      "type" 
          
       
    
  

Managed Identity for Azure Storage (2025)

Azure Table Storage

Now supports system-assigned and user-assigned managed identity:

{
  "name": "AzureTableStorageLinkedService",
  "type": "AzureTableStorage",
  "typeProperties": {
    "serviceEndpoint": "https://mystorageaccount.table.core.windows.net",
    "authenticationType": "ManagedIdentity"  // New in 2025
    // Or user-assigned:
    // "credential": {
    //   "referenceName": "UserAssignedManagedIdentity"
    // }
  }
}

Azure Files

Now supports managed identity authentication:

{
  "name": "AzureFilesLinkedService",
  "type": "AzureFileStorage",
  "typeProperties": {
    "fileShare": "myshare",
    "accountName": "mystorageaccount",
    "authenticationType": "ManagedIdentity"  // New in 2025
  }
}

Mapping Data Flows - Spark 3.3

Spark 3.3 now powers Mapping Data Flows:

Performance Improvements:

• 30% faster data processing
• Improved memory management
• Better partition handling
• Enhanced join performance

New Features:

• Adaptive Query Execution (AQE)
• Dynamic partition pruning
• Improved caching
• Better column statistics

{
  "name": "DataFlow1",
  "type": "MappingDataFlow",
  "typeProperties": {
    "sources": [
      {
        "dataset": { "referenceName": "SourceDataset" }
      }
    ],
    "transformations": [
      {
        "name": "Transform1"
      }
    ],
    "sinks": [
      {
        "dataset": { "referenceName": "SinkDataset" }
      }
    ]
  }
}

Azure DevOps Server 2022 Support

Git integration now supports on-premises Azure DevOps Server 2022:

{
  "name": "DataFactory",
  "properties": {
    "repoConfiguration": {
      "type": "AzureDevOpsGit",
      "accountName": "on-prem-ado-server",
      "projectName": "MyProject",
      "repositoryName": "adf-repo",
      "collaborationBranch": "main",
      "rootFolder": "/",
      "hostName": "https://ado-server.company.com"  // On-premises server
    }
  }
}

🔐 Managed Identity 2025 Best Practices

User-Assigned vs System-Assigned Managed Identity

System-Assigned Managed Identity:

{
  "type": "AzureBlobStorage",
  "typeProperties": {
    "serviceEndpoint": "https://mystorageaccount.blob.core.windows.net",
    "accountKind": "StorageV2"
    // ✅ Uses Data Factory's system-assigned identity automatically
  }
}

User-Assigned Managed Identity (NEW 2025):

{
  "type": "AzureBlobStorage",
  "typeProperties": {
    "serviceEndpoint": "https://mystorageaccount.blob.core.windows.net",
    "accountKind": "StorageV2",
    "credential": {
      "referenceName": "UserAssignedManagedIdentityCredential",
      "type": "CredentialReference"
    }
  }
}

When to Use User-Assigned:

• ✅ Sharing identity across multiple data factories
• ✅ Complex multi-environment setups
• ✅ Granular permission management
• ✅ Identity lifecycle independent of data factory

Credential Consolidation (NEW 2025):

ADF now supports a centralized Credentials feature:

{
  "name": "ManagedIdentityCredential",
  "type": "Microsoft.DataFactory/factories/credentials",
  "properties": {
    "type": "ManagedIdentity",
    "typeProperties": {
      "resourceId": "/subscriptions/{sub}/resourceGroups/{rg}/providers/Microsoft.ManagedIdentity/userAssignedIdentities/{identity-name}"
    }
  }
}

Benefits:

• ✅ Consolidate all Microsoft Entra ID-based credentials in one place
• ✅ Reuse credentials across multiple linked services
• ✅ Centralized permission management
• ✅ Easier audit and compliance tracking

MFA Enforcement Compatibility (October 2025)

🚨 IMPORTANT: Azure requires MFA for all users by October 2025

Impact on ADF:

• ✅ Managed identities are UNAFFECTED - No MFA required for service accounts
• ✅ Continue using system-assigned and user-assigned identities without changes
• ❌ Interactive user logins affected - Personal Azure AD accounts need MFA
• ✅ Service principals with certificate auth - Recommended alternative to secrets

Best Practice:

{
  "type": "AzureSqlDatabase",
  "typeProperties": {
    "server": "myserver.database.windows.net",
    "database": "mydb",
    "authenticationType": "SystemAssignedManagedIdentity"
    // ✅ No MFA needed, no secret rotation, passwordless
  }
}

Principle of Least Privilege (2025)

Storage Blob Data Roles:

• Storage Blob Data Reader - Read-only access (source)
• Storage Blob Data Contributor - Read/write access (sink)
• ❌ Avoid Storage Blob Data Owner unless needed

SQL Database Roles:

-- Create contained database user for managed identity
CREATE USER [datafactory-name] FROM EXTERNAL PROVIDER;

-- Grant minimal required permissions
ALTER ROLE db_datareader ADD MEMBER [datafactory-name];
ALTER ROLE db_datawriter ADD MEMBER [datafactory-name];

-- ❌ Avoid db_owner unless truly needed

Key Vault Access Policies:

{
  "permissions": {
    "secrets": ["Get"]  // ✅ Only Get permission needed
    // ❌ Don't grant List, Set, Delete unless required
  }
}

Best Practices (2025)

1. Use Databricks Job Activity (MANDATORY):

   • ❌ STOP using Notebook, Python, JAR activities
   • ✅ Migrate to DatabricksJob activity immediately
   • ✅ Define workflows in Databricks workspace
   • ✅ Leverage serverless compute (no cluster config needed)
   • ✅ Utilize advanced features (Run As, Task Values, If/Else, Repair Runs)

2. Managed Identity Authentication (MANDATORY 2025):

   • ✅ Use managed identities for ALL Azure resources
   • ✅ Prefer system-assigned for simple scenarios
   • ✅ Use user-assigned for shared identity needs
   • ✅ Leverage Credentials feature for consolidation
   • ✅ MFA-compliant for October 2025 enforcement
   • ❌ Avoid access keys and connection strings
   • ✅ Store any remaining secrets in Key Vault

3. Monitor Job Execution:

   • Track Databricks Job run IDs from ADF output
   • Log Job parameters for auditability
   • Set up alerts for job failures
   • Use Databricks job-level monitoring
   • Leverage built-in lineage tracking

4. Optimize Spark 3.3 Usage (Data Flows):

   • Enable Adaptive Query Execution (AQE)
   • Use appropriate partition counts (4-8 per core)
   • Monitor execution plans in Databricks
   • Use broadcast joins for small dimensions
   • Implement dynamic partition pruning

Resources

• Databricks Job Activity
• ADF Connectors
• Managed Identity Authentication
• Mapping Data Flows