Use when the user is working with SAP Databricks — setting up the account, administering workspaces, managing identity (users/groups/service principals/SCIM), configuring networking/security (IP lists, serverless egress), building notebooks, working with AI/ML (Genie, AI Functions, model serving, vector search), tracking billing/budget policies, sharing data from Databricks to BDC or from BDC into Databricks, or using the REST API. Trigger phrases include "SAP Databricks", "Databricks workspace", "Databricks admin", "Databricks identity", "SCIM Databricks", "Databricks notebook", "Databricks billing", "serverless budget", "Databricks API", "share to BDC", "publish to BDC Databricks", "mount BDC data".
Use when the user is working with SAP Databricks — setting up the account, administering workspaces, managing identity (users/groups/service principals/SCIM), configuring networking/security (IP lists, serverless egress), building notebooks, working with AI/ML (Genie, AI Functions, model serving, vector search), tracking billing/budget policies, sharing data from Databricks to BDC or from BDC into Databricks, or using the REST API. Trigger phrases include "SAP Databricks", "Databricks workspace", "Databricks admin", "Databricks identity", "SCIM Databricks", "Databricks notebook", "Databricks billing", "serverless budget", "Databricks API", "share to BDC", "publish to BDC Databricks", "mount BDC data".
SAP Databricks
Overview
SAP Databricks is a special edition of Databricks available as an application within SAP Business Data Cloud. It provides advanced AI and analytics capabilities fully integrated with your SAP landscape. Key benefits include pre-configured Delta Sharing access to BDC data products without replication, ability to combine SAP data with external data and publish results back to BDC, fully managed serverless compute and storage, and instant provisioning from SAP for Me with integrated SSO.
Key use cases: automated forecasting (time-series with automatic algorithm/hyperparameter selection), fine-tuning LLMs with your data, exploratory data analysis with collaborative multi-language notebooks, business intelligence with SQL editor and visualizations.
Supported regions: AWS (ap-northeast-1, ap-southeast-1, ap-southeast-2, ca-central-1, eu-central-1, us-east-1), GCP (asia-south1, europe-west3, us-central1), Azure (eastus, westeurope, westus2). Accounts can only deploy workspaces in a single region.
Setup & Accounts
Provisioning from SAP for Me
Follow SAP documentation to provision SAP Databricks from the SAP for Me portal.
Verify email and agree to terms in the Databricks sign-up flow.
The first user added becomes the initial account admin and receives an email when ready.
A first workspace is automatically created.
Initial Configuration
(Optional) Configure network settings for enhanced security.
Add users and groups manually or via SAP Cloud Identity Service Identity Provisioning (SCIM integration).
Users log in at login.databricks.com with SSO credentials.
Creating Additional Workspaces
From the account console Workspaces tab:
Click Create workspace.
Enter workspace name and select region.
Click Create.
Assign users/groups permissions on the Permissions tab using Add permissions.
Note: Accounts can only deploy workspaces in a single region.
Admin Roles & Responsibilities
Account Admin
Create and manage workspaces
Configure networking controls and IP access lists
Manage users, groups, and service principals at account level
Assign metastore privileges and admin roles
Access account console at accounts.cloud.databricks.com or via workspace selector → Manage account
Transfer account owner role (contact Databricks support if needed)
Workspace Admin
Manage access control for workspace objects (notebooks, queries)
Create and manage serverless SQL warehouses
Manage workspace identities and admin settings
Access system tables and audit logs
Create external location connections to cloud storage (S3, Azure, GCS)
Access workspace settings via username menu → Settings
Create job and schedule via Schedule a notebook if not already assigned.
Git Folders
Visual Git client and API in Databricks.
Support cloning, committing, pushing, pulling, branch management, diff comparison.
Develop notebooks/files with Git version control for CI/CD.
Genie Code
Context-aware AI assistant for data and code.
Available in SQL editor and notebooks.
Features: AI autocomplete, natural language data filtering, error diagnosis, quick-fix suggestions.
Web Terminal
Interactive shell command execution (if enabled by account admin).
Useful for batch operations on multiple files.
Launch from notebook (serverless compute environment 2) via terminal icon in right sidebar.
AI & Machine Learning
Mosaic AI Platform
Unifies AI lifecycle from data collection/preparation to model development/LLMOps to serving/monitoring.
Features Available
AI Playground: Chat-like environment to test, prompt, and compare supported LLMs.
AI Functions: Built-in functions to apply AI (text translation, sentiment analysis, etc.) on Databricks data. Run from notebook or SQL editor.
Mosaic AI Gateway: Centralized service for governing, monitoring, and managing access to generative AI models and model serving endpoints. Provides governance, monitoring, production readiness, secure traffic management.
Mosaic AI Model Serving: Deploy, govern, and query AI models for real-time/batch inference. Each served model available as REST API. Configure for generative AI (foundation models, third-party models).
Mosaic AI Vector Search: Vector database for embedding vectors, automatic knowledge base sync. Built-in to Databricks, integrated with governance and productivity tools. Use for RAG, recommendation systems, image recognition.
Lakehouse Monitoring: Monitor statistical properties and data quality across tables. Track ML model performance and drift via inference tables with automatic payload logging.
Managed MLflow: Open-source AI engineering platform for agents, LLMs, ML models. Debug, evaluate, monitor, optimize production AI apps. Experiment tracking, evaluation, model registry, deployment.
Mosaic AI Agent Framework: Tools for building, deploying, evaluating production-quality agents (RAG apps). Compatible with LangChain, LlamaIndex. Leverage Databricks managed Unity Catalog and Agent Evaluation.
Mosaic AI Agent Evaluation: Evaluate quality, cost, latency of agentic AI applications (RAG, chains). Identify issues and root causes across development, staging, production phases. Metrics logged to MLflow Runs.
AutoML Forecasting: Automatically select best algorithm and hyperparameters for time-series data on fully-managed compute.
Foundation Model Fine-tuning: Customize foundation models with your data. Requires supported region (us-east-1). Less data, time, compute than training from scratch.
All data assets and ML artifacts (models, functions) discoverable and governed in single catalog. Track lineage from raw data to production model. Built-in monitoring saves quality metrics to tables.
External Data & Delta Sharing
Connect to Cloud Storage (Read-Only)
S3 (AWS): Create external locations to AWS S3 bucket (admin only). Must be read-only to prevent data loss.
Azure Blob Storage: Create external locations to Azure blob storage/Data Lake Storage (admin only). Must be read-only.
GCS (GCP): Create external locations to Google Cloud Storage bucket (admin only). Must be read-only.
Delta Sharing
Receive Delta Sharing shares from workspaces inside/outside account.
SAP Databricks accounts are recipients only; cannot initiate shares outside SAP.
Publishing to BDC
Pre-configured Delta Sharing to BDC
Each workspace pre-populated with Delta Sharing recipient sap-business-data-cloud.
Allows sharing data from SAP Databricks back to SAP BDC.
Create & Publish a Share
Create share via Catalog Explorer, SQL, or CLI.
Add data to share.
Click Share data.
Semantic Metadata
Enrich shared data for discovery in SAP BDC.
Use SAP BDC SDK to describe data using CSN and ORD.
Receiving SAP Data into Databricks
Delta Sharing from SAP BDC
Each SAP application automatically appears as provider in SAP Databricks account.
Activating data product in SAP BDC automatically creates Delta Share to SAP Databricks.
Data products mounted to catalogs available in notebooks, SQL editor, AI/ML products.
Make SAP Data Available
Requires USE PROVIDER and CREATE CATALOG metastore permissions.
From Catalog Explorer:
Click Catalog.
Click Delta Sharing.
Click SAP provider.
Click Mount to catalog for target table.
Create new catalog or mount to existing.
SAP Semantic Metadata
Automatically ingested into Unity Catalog at table level when accessed.
Metadata synced from SAP BDC includes:
Table/column comments: Purpose descriptions.
Primary keys: Synced as Unity Catalog primary key constraints.
Foreign keys: Relationships within same share (cross-share not supported).
SAP governance tags: System tags in sap.PersonalData.* namespace classifying personal/sensitive data.
SAP Governance Tags (Read-Only)
Tags in sap.PersonalData namespace synced as system governed tags.
Do not manually assign/modify/delete tags in sap.* namespace (system-reserved).
