Skip to content
DP-700 Fabric Engineer Hub

Exam DP-700: Comprehensive Data Engineering Glossary

Exam DP-700: Comprehensive Data Engineering Glossary

Source

  • Provider: Microsoft
  • Platform: Microsoft Learn
  • Target Exam: Exam DP-700
  • Content type: Technical Glossary
  • Date captured: 2026-08-15
  • Last reviewed: 2026-08-15

A

  • ACID Transactions: Atomicity, Consistency, Isolation, and Durability guarantees provided on Delta Lake tables via the _delta_log/ transaction journal.
  • Adaptive Query Execution (AQE): Spark runtime optimization engine that dynamically re-optimizes query plans based on stage statistics (e.g., coalescing shuffle partitions, handling data skew).
  • Apache Airflow: Open-source workflow orchestration tool natively hosted as a managed Fabric workspace item for Python DAG-based scheduling.

B

  • Broadcast Join: Spark join strategy where a small dimension DataFrame ($<100$ MB) is copied to all worker nodes to avoid expensive full shuffle data exchanges across the network.
  • Bronze Layer: The raw landing zone in a Medallion Architecture where incoming data is ingested append-only without schema cleansing.
  • Bursting: Fabric capacity mechanism allowing compute workloads to temporarily consume compute units (CUs) above baseline SKU capacity without immediate throttling.

C

  • Capacity Unit (CU): Standard measurement of computing power in Microsoft Fabric (e.g., F64 = 64 CUs).
  • Change Data Capture (CDC): Software pattern that identifies and captures real-time changes (inserts, updates, deletes) made to database tables.
  • Column-Level Security (CLS): Security control in Fabric Warehouse & SQL endpoints that restricts unauthorized users from viewing specific column values.

D

  • Dataflow Gen2: Low-code data preparation engine built on Power Query Online that supports outputting directly to Lakehouses and Warehouses with Fast Copy acceleration.
  • Data Factory Pipeline: Cloud-scale control flow orchestration engine for executing Copy activities, Notebooks, and Stored Procedures.
  • Delta Lake: Open-source storage layer that brings ACID transactions and Parquet columnar compression to OneLake.
  • Direct Lake Mode: Power BI query engine mode that reads Delta Lake parquet files directly from OneLake without loading data into an in-memory VertiPaq model or issuing slow DirectQuery SQL queries.

E

  • Eventhouse: Scalable Fabric container optimized for real-time analytics, time-series events, and log streaming.
  • Eventstream: Fabric streaming ingestion component that captures, transforms (filters/aggregations), and routes high-frequency data streams.

K

  • KQL (Kusto Query Language): Fast, expressive, read-only query language optimized for exploring large volumes of structured, semi-structured, and time-series streaming data.
  • KQL Update Policy: Automated stream transformation trigger attached to KQL tables that runs a function on newly ingested data to populate a curated target table.

L

  • Liquid Clustering: Modern Delta Lake feature that replaces static hive-style partitioning and Z-Ordering with flexible, background clustering columns.

M

  • Medallion Architecture: Data design pattern organizing data into Bronze (raw), Silver (cleansed/conformed), and Gold (curated business star schema) tiers.
  • Mirroring: Continuous, near-real-time replication of operational relational and NoSQL databases into Delta Lake tables in OneLake.

O

  • OneLake: Single, unified SaaS data lake for an entire Microsoft Fabric tenant, built on ADLS Gen2.
  • OneLake Shortcut: Virtualized pointer to internal or external data storage (ADLS Gen2, AWS S3, Google Cloud Storage, Dataverse) without moving or copying physical files.
  • OPTIMIZE: Delta Lake command that compacts small files into optimal 1GB parquet files and applies sorting.

R

  • Row-Level Security (RLS): Security predicate filter applied in SQL endpoints to restrict rows returned based on the executing user’s Entra ID identity.

S

  • Silver Layer: The cleansed, deduplicated, and normalized tier of a Medallion Architecture.
  • Smoothing: Capacity mechanism that averages out interactive operations over 5 minutes and background batch operations over 24 hours to prevent unnecessary throttling.

V

  • VACUUM: Delta Lake maintenance operation that purges deleted historical files older than the retention threshold (default 7 days).
  • V-Order: Microsoft-proprietary write-time sorting algorithm applied to Delta parquet files to optimize scan performance for Power BI Direct Lake and SQL engines.