Final Revision: Scenario and Component Guide

Match each scenario to its best starting choice.

Component Selection - Move large volumes of data with minimal transformation: Copy Data activity - Apply low-code Power Query transformations: Dataflow Gen2 - Coordinate multiple activities with dependencies and schedules: Data pipeline - Perform complex large-scale transformations: Spark notebook - Store raw and curated Delta tables: Lakehouse - Build a relational star schema with full T-SQL DML: Fabric Data Warehouse - Ingest and route streaming events: Eventstream - Store and query high-velocity time-series data: Eventhouse / KQL database - Show auto-refreshing operational visuals: Real-Time Dashboard - Trigger actions when a live condition is met: Activator - Read OneLake data directly through Power BI: Direct Lake semantic model

Commonly confused concept pairs and the key distinction between them.

Concept Comparisons - Dataflow Gen2 vs Data pipeline: transformation worker vs orchestration manager - Copy Data vs Dataflow Gen2: high-speed movement vs visual transformation - Lakehouse vs Warehouse: Spark, files and Delta tables vs relational SQL-led warehouse - SQL analytics endpoint vs Warehouse: read-only lakehouse SQL vs read-write transactional T-SQL - Eventstream vs Eventhouse: ingest, transform and route in motion vs store and query real-time data - Pipeline parameter vs Pipeline variable: runtime input vs value stored or changed during a run - Optimize Write vs OPTIMIZE: prevent small files during writes vs compact existing files - V-Order vs Partitioning: file-level read optimisation vs physical folder layout - Workspace role vs Item permission: broad workspace collaboration vs targeted item access - Git integration vs Deployment pipeline: source control synchronisation vs environment promotion

Use these checklists for a last pass before the exam.

Delta and Lakehouse - A Delta table equals Parquet data files plus a _delta_log transaction history - ACID, schema enforcement, CRUD and time travel are core Delta benefits - Use OPTIMIZE for existing small files and VACUUM for obsolete files - Use partitioning selectively on very large tables with sensible filter columns - The lakehouse SQL analytics endpoint is read-only for table data

Warehouse - Facts hold measurements; dimensions provide context - Load dimensions before facts when facts require surrogate-key lookups - Type 1 SCD overwrites; Type 2 inserts a new historical row - Use full loads for simplicity and incremental loads for efficiency

Real-Time and Security - Eventstream handles in-flight ingestion, transformation and routing - Eventhouse stores high-velocity time-series data in KQL databases - Apply least privilege at workspace, item, compute and data layers - RLS filters rows, CLS blocks columns, and masking obscures displayed values - DENY always overrides GRANT