Unity Catalog: The Three-Level Namespace

What is Unity Catalog?

The three-level namespace

Every data object in Unity Catalog has a fully qualified name:

catalog.schema.object

Level	Purpose	Analogy	Example
Catalog	Top-level container — typically one per business domain or environment	Floor in a building	sales, finance, dev_sandbox
Schema (database)	Groups related objects within a catalog	Room on a floor	sales.raw, sales.curated, sales.reports
Object	The actual data — tables, views, volumes, functions	File cabinet in a room	sales.curated.daily_revenue

-- Fully qualified reference
SELECT * FROM sales.curated.daily_revenue;

-- Set default catalog and schema to avoid typing the full path
USE CATALOG sales;
USE SCHEMA curated;
SELECT * FROM daily_revenue;  -- now just the table name

Naming conventions

The exam tests your ability to design naming conventions for different requirements:

By environment (isolation)

Pattern	Example	Use Case
Separate catalogs per environment	dev_sales, staging_sales, prod_sales	Full data isolation between dev/staging/prod
Separate schemas per environment	sales.dev_raw, sales.staging_raw, sales.prod_raw	Shared catalog, environment isolation at schema level
Separate catalogs per team	data_engineering, data_science, analytics	Team-based isolation

Dr. Sarah Okafor at Athena Group chooses separate catalogs per environment (dev, staging, prod) because her security team requires complete isolation — developers must never accidentally query production data.

By external sharing

When sharing data with external partners, create a dedicated sharing catalog:

-- Catalog specifically for data shared via Delta Sharing
CREATE CATALOG IF NOT EXISTS shared_external;
CREATE SCHEMA IF NOT EXISTS shared_external.partner_freshmart;

Exam tip: Naming conventions are tested via scenarios. If the question mentions “isolation” or “prevent accidental access” — think separate catalogs. If it mentions “sharing” — think dedicated sharing catalog.

ℹ️ Real-world naming patterns

Common patterns seen in production:

Pattern	Structure	Pros	Cons
env_domain	prod_sales.raw.orders	Clear environment + domain	Lots of catalogs
domain with env schemas	sales.prod_raw.orders	Fewer catalogs	Less isolation
domain with env catalogs	prod.sales.orders	Clean, hierarchical	Requires strict permissions

Most enterprise teams use environment-first (prod_sales, dev_sales) because Unity Catalog permissions are inherited — setting permissions at the catalog level propagates down to all schemas and tables.

Creating catalogs

-- Create a catalog
CREATE CATALOG IF NOT EXISTS prod_sales
  COMMENT 'Production sales data for all regions';

-- View all catalogs
SHOW CATALOGS;

-- Describe a catalog
DESCRIBE CATALOG prod_sales;

A catalog is bound to a storage location — where its managed tables physically store data. By default, catalogs use the metastore’s root storage. You can override this:

-- Catalog with custom storage
CREATE CATALOG prod_sales
  MANAGED LOCATION 'abfss://sales-container@adlsaccount.dfs.core.windows.net/prod';

Creating schemas

-- Create a schema within a catalog
CREATE SCHEMA IF NOT EXISTS prod_sales.raw
  COMMENT 'Raw ingested data before any transformation';

CREATE SCHEMA IF NOT EXISTS prod_sales.curated
  COMMENT 'Cleaned, validated, and conformed data';

CREATE SCHEMA IF NOT EXISTS prod_sales.aggregated
  COMMENT 'Business-level aggregates for reporting';

Schemas can also have a managed location that overrides the catalog’s default:

CREATE SCHEMA prod_sales.sensitive
  MANAGED LOCATION 'abfss://secure-container@adlsaccount.dfs.core.windows.net/sensitive';

Volumes: the file storage layer

Volumes are Unity Catalog’s way of managing files (not tables). Think of volumes as governed file directories:

Volume Type	Description	Use Case
Managed volume	Files stored in Unity Catalog’s managed storage	Internal data files, staging area
External volume	Points to existing ADLS/S3 location	Landing zone for incoming data files

-- Create a managed volume
CREATE VOLUME IF NOT EXISTS prod_sales.raw.landing_files
  COMMENT 'Landing zone for CSV/JSON files from partners';

-- Create an external volume pointing to existing storage
CREATE EXTERNAL VOLUME prod_sales.raw.partner_uploads
  LOCATION 'abfss://partner-landing@adlsaccount.dfs.core.windows.net/uploads';

Ravi uses volumes at DataPulse to manage the CSV files that clients upload before they’re ingested into Delta tables.

💡 Volumes vs. tables: when to use which

Use	Tables	Volumes
Structured data (rows and columns)	✅	❌
Files (CSV, JSON, images, PDFs)	❌	✅
Queryable with SQL	✅	Read with read_files() or COPY INTO
Schema enforcement	✅ (Delta)	❌ (files are as-is)
Governed by Unity Catalog	✅	✅

Exam tip: Volumes are for files, tables are for data. If the question mentions “landing zone for raw files” or “file-based ingestion” — think volumes.

Question

What are the three levels of the Unity Catalog namespace?

Click or press Enter to reveal answer

Answer

Catalog > Schema > Object. Fully qualified name: catalog.schema.table_name. Catalogs group by domain/environment, schemas group related objects, objects are the actual tables/views/volumes.

Click to flip back

Question

When should you use separate catalogs per environment vs. separate schemas?

Click or press Enter to reveal answer

Answer

Separate catalogs (dev_sales, prod_sales) for full isolation — permissions inherit at catalog level and prevent accidental cross-environment access. Separate schemas (sales.dev_raw, sales.prod_raw) when teams need shared catalog-level resources but environment separation at the data level.

Click to flip back

Question

What is a volume in Unity Catalog?

Click or press Enter to reveal answer

Answer

A governed file storage object. Managed volumes store files in UC's managed storage. External volumes point to existing ADLS locations. Use volumes for files (CSV, JSON, images) — use tables for structured data.

Click to flip back

Question

What is the naming convention for data shared externally via Delta Sharing?

Click or press Enter to reveal answer

Answer

Create a dedicated sharing catalog (e.g., shared_external) with schemas per partner. This isolates shared data from internal catalogs and makes permission management clear.

Click to flip back

🎬 Video coming soon

Knowledge check

Knowledge Check

Dr. Sarah Okafor is designing Athena Group's Unity Catalog structure. She needs to ensure developers can never accidentally query or modify production data. Which naming strategy should she recommend?

AOne catalog with schemas per environment (sales.dev_raw, sales.prod_raw)BSeparate catalogs per environment (dev_sales, staging_sales, prod_sales)COne catalog with table-name prefixes (dev_orders, prod_orders)DSeparate workspaces with no shared metastore

Check Answer

Knowledge Check

Mei Lin needs to set up a landing zone at Freshmart where external suppliers upload CSV files before they're ingested into Delta tables. The files should be governed by Unity Catalog. What should she create?

AA managed table with CSV formatBA schema called 'landing_zone'CAn external volume pointing to the supplier upload locationDA foreign catalog connected to the supplier's system

Check Answer

Knowledge Check

Tomás is creating the data catalog structure for NovaPay's fraud detection system. He has three data layers: raw transactions, enriched data, and fraud alerts. Which Unity Catalog structure follows best practices?

AThree catalogs: raw_transactions, enriched_data, fraud_alertsBOne catalog (fraud_detection) with three schemas: raw, enriched, alertsCOne schema with table prefixes: raw_txn, enriched_txn, alert_fraudDThree volumes: raw, enriched, alerts

Check Answer

---

Next up: Tables, Views & External Catalogs — managed vs external tables, views, materialized views, foreign catalogs, DDL, and AI/BI Genie.