Create Tables, Views, and Materialized Views in Unity Catalog
Understanding Object Creation in Unity Catalog
Within Azure Databricks, Unity Catalog provides centralized governance for data and AI assets. Data engineers use catalogs and schemas to organize tables, views, and materialized views logically across environments. This structure improves security, discoverability, collaboration, and administrative consistency. The DP-750 exam expects candidates to understand how these objects behave and when each object type should be used.
Tables represent the primary storage object for structured data. Most Azure Databricks workloads use Delta tables because Delta Lake supports ACID transactions, schema evolution, time travel, and scalable analytics processing. Engineers create managed tables when Databricks should control the storage lifecycle. External tables are more appropriate when organizations already maintain governed cloud storage locations externally.
The following example creates a managed Delta table.
CREATE TABLE finance.transactions (
transaction_id BIGINT,
customer_id BIGINT,
transaction_date DATE,
amount DECIMAL(12,2)
);
An external table references existing storage rather than using Databricks-managed storage.
CREATE TABLE finance.external_transactions
USING DELTA
LOCATION 'abfss://finance@storageaccount.dfs.core.windows.net/transactions/';
Understanding storage management responsibilities is important for the DP-750 exam because governance and architecture decisions frequently depend on these differences.
Using Views to Simplify Data Access
Views provide logical abstractions over underlying tables. A view stores a SQL query definition instead of physically storing data. Engineers create views to simplify complex joins, restrict sensitive columns, standardize business logic, or improve usability for reporting teams.
Views are especially useful within governed enterprise environments because analysts can consume curated datasets without direct access to raw ingestion tables. This approach supports security and consistency across reporting platforms.
The following example creates a view that filters high-value transactions.
CREATE VIEW finance.high_value_transactions AS
SELECT transaction_id, customer_id, amount
FROM finance.transactions
WHERE amount > 5000;
Views also help organizations implement medallion architecture patterns. Bronze tables may contain raw source data, while Silver and Gold views expose cleansed and business-friendly representations. Engineers should understand that standard views do not improve performance directly because the underlying query executes each time the view is queried.
Unity Catalog permissions also affect view creation. Engineers require USE CATALOG, USE SCHEMA, and CREATE VIEW privileges within the target schema. Governance and access control remain important exam objectives.
Improving Performance with Materialized Views
Materialized views differ from standard views because they physically store query results. This capability improves performance for expensive analytical workloads and frequently accessed reporting queries. Materialized views reduce repeated computation because Databricks refreshes stored results automatically or on a defined schedule.
Materialized views are particularly valuable when dashboards repeatedly execute large aggregations against massive datasets. Instead of recalculating the same query repeatedly, users query precomputed results.
The following example creates a materialized view.
CREATE MATERIALIZED VIEW finance.monthly_sales_summary AS
SELECT
date_trunc('month', transaction_date) AS sales_month,
SUM(amount) AS total_sales
FROM finance.transactions
GROUP BY date_trunc('month', transaction_date);
Engineers must balance performance gains against storage costs and refresh overhead. Materialized views consume storage because they persist query results physically. The DP-750 exam often focuses on selecting the correct object based on workload requirements, governance expectations, and query performance considerations.
Governance, Organization, and Operational Considerations
Unity Catalog supports centralized governance across tables, views, and materialized views. Administrators can assign permissions at catalog, schema, or object level. This governance model simplifies enterprise security management and improves auditing capabilities.
Naming conventions are also important. Many organizations separate environments into development, testing, and production catalogs. Within schemas, data may be divided into Bronze, Silver, and Gold layers to support ingestion, transformation, and reporting workloads.
Lineage capabilities within Unity Catalog help engineers understand dependencies between notebooks, pipelines, tables, and downstream reporting objects. This visibility improves troubleshooting and impact analysis when schema changes occur.
For the DP-750 exam, candidates should understand the differences between managed and external tables, the purpose of views, the performance benefits of materialized views, and the governance capabilities provided by Unity Catalog.
Links
Microsoft Certified: Azure Databricks Data Engineer Associate – Certifications | Microsoft Learn
Tables and views in Azure Databricks – Azure Databricks | Microsoft Learn
What is a view? – Azure Databricks | Microsoft Learn
Use materialized views in Databricks SQL – Azure Databricks | Azure Docs
Example DP-750 Style Questions
Question 1
A retail company needs to store transactional sales data with support for ACID transactions and schema enforcement. Which Unity Catalog object should the engineer create?
Question 2
An organization already maintains Delta files within Azure Data Lake Storage Gen2. The engineering team must register the data in Unity Catalog without relocating the files. Which type of table should the engineer use?
Question 3
A business analyst requires access to filtered customer information without direct access to sensitive raw tables. Which Unity Catalog object best satisfies this requirement?
Question 4
A reporting dashboard repeatedly executes a costly aggregation query against billions of records. Query performance has become unacceptable during business hours. Which object would most likely improve performance?
Question 5
An engineer successfully accesses a catalog but cannot create a view inside a schema. Which additional permissions are most likely required?
Question 6
A company organizes data into Bronze, Silver, and Gold layers. Which Unity Catalog feature helps logically separate these layers?
Question 7
A data engineer wants to identify downstream reports and notebooks dependent on a specific Delta table. Which Unity Catalog capability supports this requirement?
Question 8
A team currently uses standard SQL views for large reporting workloads. Users complain about slow query performance. Why might materialized views provide a better solution?
Answers
Answer 1
The engineer should create a Delta table.
Answer 2
The engineer should create an external table.
Answer 3
The engineer should create a view.
Answer 4
The engineer should create a materialized view.
Answer 5
The engineer likely requires USE SCHEMA and CREATE VIEW permissions.
Answer 6
Schemas help logically organize Bronze, Silver, and Gold layers.
Answer 7
Unity Catalog lineage tracking supports dependency analysis.
Answer 8
Materialized views store precomputed query results, reducing repeated computation and improving reporting performance.