Databricks Lakehouse Platform Administrator Study Strategy¶

Study Approach¶

Phase 1: Platform Fundamentals (Week 1)¶

1. Account vs Workspace - Account console, workspace console, admin roles
2. Identity Management - Users, groups, service principals, entitlements
3. SCIM and SSO - Automated provisioning, SAML/OIDC configuration
4. 📖 Administration Guide
5. 📖 Identity Management

Phase 2: Compute and Data (Week 2)¶

1. Clusters - Configuration, auto-scaling, spot instances, instance pools
2. Cluster Policies - Governance, cost control, security enforcement
3. SQL Warehouses - Types, sizing, scaling, permissions
4. Unity Catalog - Hierarchy, managed vs external tables, volumes, grants
5. 📖 Compute
6. 📖 Unity Catalog

Phase 3: Security and Exam Prep (Week 3)¶

1. Network Security - VPC injection, Private Link, IP access lists
2. Encryption - Default vs customer-managed keys, secrets
3. Audit Logging - Log delivery, system tables, compliance
4. Practice exams and review
5. 📖 Security Guide
6. 📖 Audit Logs

Recommended Resources¶

Primary Resources¶

• Databricks Academy - Official training courses
• Databricks Certification Page - Exam details
• Databricks Documentation - Full documentation
• Databricks Community Forum - Discussion and Q&A

Documentation to Focus On¶

• 📖 Admin Guide - Most exam-relevant section
• 📖 Security Best Practices - Security configuration
• 📖 Unity Catalog - Data governance
• 📖 Cluster Policies - Compute governance

Exam Tactics¶

Keywords to Watch For¶

• "Cost control" or "limit compute" - Cluster policies
• "Automate user management" - SCIM provisioning
• "Non-human identity" or "CI/CD" - Service principal
• "No public internet" - Private Link
• "Own encryption keys" - Customer-managed keys (CMK)
• "Track data access" - Audit logs + Unity Catalog
• "Share data externally" - Delta Sharing
• "Cloud storage access" - External locations + storage credentials
• "Query concurrency" - SQL warehouse scaling (clusters)
• "Restrict workspace access" - IP access lists
• "Account-level vs workspace-level" - Centralized (account) vs per-workspace

Common Pitfalls¶

• Account-level SCIM is preferred over workspace-level SCIM
• Cluster policies restrict what users can configure - they do not create clusters
• Removing "Allow cluster creation" entitlement does not block policy-based cluster creation
• External table DROP removes metadata only, not data
• Managed table DROP removes both metadata and data
• SQL warehouse size controls single-query performance; cluster count controls concurrency
• Service principals should use OAuth, not PATs
• IP access lists restrict UI/API access, not cluster-to-data-source traffic
• Unity Catalog permissions require USE CATALOG and USE SCHEMA before table-level grants work

Time Management¶

• 90 minutes for 45 questions - 2 minutes per question
• Workspace admin and Unity Catalog questions are most frequent (55% combined)
• Security questions are fewer but detail-oriented
• Eliminate obvious wrong answers first

Readiness Indicators¶

• You understand account console vs workspace console responsibilities
• You can configure SCIM and SSO for identity management
• You know how to write cluster policies in JSON
• You understand Unity Catalog hierarchy and permission model
• You can explain managed vs external tables and volumes
• You know network security options (VPC injection, Private Link, IP lists)
• You understand customer-managed key configuration
• You can query system tables for audit information
• You score 70%+ consistently on practice questions