Four domains — the two data domains carry the most weight (30–35% each), with environment setup and Unity Catalog governance at 15–20% each. Every summary below is paraphrased from the official skills outline; the bullet-level objectives in Azure Mastery are tagged so you always know which domain you're being tested on.
Set up and configure an Azure Databricks environment15–20%
The platform-setup domain. Selecting and configuring compute — job, serverless, warehouse, classic, and shared — with performance settings (CPU, node count, autoscaling, termination, pooling) and feature settings (Photon acceleration, runtime/Spark version, machine learning); installing libraries and setting compute access permissions; and creating and organising Unity Catalog objects — catalogs, schemas, volumes, tables, views, and materialized views — plus foreign catalogs, DDL on managed and external tables, and AI/BI Genie instructions.
Secure and govern Unity Catalog objects15–20%
The governance domain. Granting privileges to principals (users, service principals, groups); implementing table-, column-, and row-level security; reading Azure Key Vault secrets and authenticating with service principals and managed identities; and governing with table and column descriptions for discovery, ABAC tags and policies, row filters and column masks, data-retention policies, data lineage in Catalog Explorer, audit logging, and a secure Delta Sharing strategy.
Prepare and process data30–35%
The hands-on heart of the exam. Modelling data in Unity Catalog — ingestion logic, table formats (Delta, Parquet, JSON, Iceberg), partitioning, SCD types, temporal history tables, and clustering (liquid clustering, Z-ordering, deletion vectors); ingesting batch and streaming data with Lakeflow Connect, Auto Loader, notebooks, CTAS, COPY INTO, CDC feeds, Spark Structured Streaming, and Azure Event Hubs; cleansing and transforming with joins, aggregation, pivoting, and merge/insert/append; and enforcing data-quality constraints with schema enforcement and pipeline expectations.
Deploy and maintain data pipelines and workloads30–35%
The operations domain. Designing pipelines and choosing between notebooks and Lakeflow Spark Declarative Pipelines; implementing Lakeflow Jobs with triggers, schedules, alerts, and automatic restarts; applying development-lifecycle practices — Git branching and pull requests, testing strategies, and Databricks Asset Bundles deployed via CLI or REST; and monitoring, troubleshooting, and optimising — cluster cost, Spark skew, spilling, shuffle issues via the DAG and Spark UI, OPTIMIZE and VACUUM, and log streaming with Azure Monitor.
