Best practices
Find guidance for the decisions you make when you build, govern, and operate Azure Databricks. Start with the Well-Architected Framework for a platform review, or choose a focused guide below.
Architecture#
- Databricks well-architected framework: Review an Azure Databricks platform against the seven architecture pillars.
- Platform lessons: Compare architecture, networking, compute, and operating decisions from platform delivery.
- Community best practices: Find vendor and practitioner guidance across architecture, governance, security, and cost.
Identity and governance#
- Identity with Entra ID and SCIM: Choose an account identity and group provisioning approach with Entra ID and SCIM.
- Unity Catalog grants: Understand privilege inheritance, discovery, and Terraform grant ownership.
- PII and ABAC governance: Plan sensitive-data controls with governed tags, masks, row filters, and groups.
Development and delivery#
- Terraform vs DABs: Separate shared infrastructure from application delivery with Terraform and bundles.
- Lakeflow data engineering: Plan ingestion, pipelines, orchestration, replay, and production ownership.
- Databricks Agent Skills: Install and use Databricks Agent Skills with Claude Code, Codex, and OpenCode.
Operations and workloads#
- Delta and storage: Choose table layouts, optimization, retention, and VACUUM safeguards.
- SQL and BI: Plan SQL warehouses, dashboards, Genie, BI access, and cost controls.
- AI and ML platform: Govern and operate model tracking, serving, evaluation, and AI workloads.
More ways to browse#
- Migration guides cover assessment, conversion, validation, and cutover.
- Browse topics to find related guidance and sources across the wiki.
- Product updates collect release notes and vendor and community sources.