Databricks Glossary: Essential Terms Every User Should Know

This glossary defines the core Databricks vocabulary, lakehouse, Delta Lake, Unity Catalog, DBU, Photon, clusters, SQL Warehouses, and more, so every user shares one accurate mental model. Knowing these terms precisely prevents the most common misunderstandings about how the platform works and what each part costs. After reading, you will recognize each essential term, understand how the terms relate, and know where to apply them.

• Define the most important Databricks terms in plain, precise language
• See how core concepts (storage, compute, governance, AI) relate to one another
• Use the terms correctly when configuring and discussing the platform

Who this is for: New Databricks users and anyone needing a quick, accurate reference for platform terminology.

Part of the What is Databricks section in the Databricks tutorial series.

Architecture / Concept Overview: Databricks Glossary: Essential Terms Every User Should Know

The essential terms cluster into four conceptual groups: storage (how data is kept), compute (how data is processed), governance (how data is controlled), and intelligence (how data drives AI). Seeing these groups helps you place any new term you encounter and understand which concern it addresses.

*The lakehouse organizes terminology into four groups: storage, compute, governance, and intelligence.*

Many terms appear together in a typical data flow, which clarifies how they relate in practice.

*Where common terms appear in a flow: Auto Loader ingests to Delta, Photon-powered compute transforms it, and a SQL Warehouse serves it.*

Key Terms

Lakehouse

An architecture combining data-warehouse reliability and performance with data-lake openness and scale on a single storage layer.

Delta Lake

The open table format adding ACID transactions, time travel, schema enforcement, and performance optimizations to files on object storage.

Unity Catalog

The unified governance layer for permissions, lineage, discovery, and auditing across data and AI assets.

DBU (Databricks Unit)

A normalized unit of processing capacity consumed over time; the basis for Databricks consumption charges.

Photon

A vectorized C++ query engine that accelerates SQL and DataFrame workloads compared to standard Spark execution.

Medallion architecture

A layered design refining data through Bronze (raw), Silver (cleaned), and Gold (curated) tables.

SQL Warehouse

Compute optimized for BI and SQL analytics on lakehouse tables, with autoscaling and Photon.

Mosaic AI

The platform's suite for building, tuning, serving, and governing machine learning and generative AI.

Prerequisites and Setup

• A Databricks workspace to try the terms in context
• Unity Catalog access to a catalog and schema
• Basic SQL familiarity
• This glossary open as a reference while you work

Step-by-Step Implementation

1. See Delta Lake in action

   Create a Delta table to make the most fundamental storage term concrete.

   -- SQL cell - create a Delta table (the lakehouse storage unit)\nCREATE TABLE main.demo.products (id INT, name STRING) USING delta;

2. Observe time travel

   Inspect a table's history to understand the "version" and "time travel" terms.

   -- SQL cell - view the transaction history of a Delta table\nDESCRIBE HISTORY main.demo.products;

3. Apply governance with Unity Catalog

   Grant access to see how governance terminology maps to real commands.

   -- SQL cell - a Unity Catalog grant\nGRANT SELECT ON TABLE main.demo.products TO `analysts`;

4. Measure consumption in DBUs

   Query billing system tables to connect the DBU term to actual usage.

   -- SQL cell - DBUs consumed recently\nSELECT sku_name, SUM(usage_quantity) AS dbus\nFROM system.billing.usage\nWHERE usage_date >= current_date() - INTERVAL 7 DAYS\nGROUP BY sku_name;

5. Run a Photon-accelerated query

   Execute an aggregate on Photon-capable compute to ground the performance terminology.

   -- SQL cell - scan-heavy query that benefits from Photon\nSELECT name, COUNT(*) AS n FROM main.demo.products GROUP BY name;

Configuration Reference

Monitoring, Cost, and Security Considerations

Monitoring

Knowing the terms makes system tables readable: system.billing.usage measures DBUs, system.query.history tracks queries, and DESCRIBE HISTORY shows Delta operations. Precise vocabulary turns raw telemetry into actionable observability.

Cost Optimisation

The DBU is the cost vocabulary you must master: lowering DBUs (via Photon, autoscaling, and auto-stop) is how you control spend. Understanding compute terms like SQL Warehouse vs job cluster directly informs cheaper choices.

Security and Governance

Governance terms, Unity Catalog, grants, row filters, column masks, and lineage, are the language of platform security. Using them correctly ensures access requests and audits are unambiguous.

Common Pitfalls and Recommended Patterns

• Confusing Delta Lake (format) with the lakehouse (architecture): they are related but distinct.
• Treating a DBU as a fixed price: it is a unit of consumption multiplied by a rate.
• Mixing up clusters and SQL Warehouses: clusters run general code, warehouses serve SQL/BI.
• Assuming Photon is automatic everywhere: confirm compute supports and enables it.
• Using "catalog" loosely: in Unity Catalog it is the top namespace level, not a vague synonym for database.

Frequently Asked Questions

Related Topics

• The Databricks Data Intelligence Platform Explained — Where these terms come together.
• What is a Data Lakehouse? — The architecture behind the core vocabulary.
• Core Components of the Databricks Platform — The components many terms name.
• Databricks Pricing and DBU (Databricks Unit) Explained — A deeper look at the DBU term.