Data Governance · Topic 3 of 7

Horizon Catalog and Data Protection

Set a data protection baseline with Snowflake Horizon Catalog, then share and federate data outside Snowflake with Iceberg REST, Marketplace and shares.

  • 4 min
In a nutshell
  • Horizon Catalog covers discovery, governance, security and interoperability for data inside and outside Snowflake.
  • Protection baseline: classify and tag, policies, audit with ACCESS_HISTORY, monitor with the Trust Center.
  • Iceberg REST Catalog API opens Snowflake managed Iceberg tables to engines such as Spark, Flink and Trino for reads and writes; policy enforcement is documented for Spark.

What it is

Horizon Catalog is Snowflake's built in catalog and governance layer. It covers data stored in Snowflake and data kept outside it, across engines, formats and clouds. Its job is to keep data protected and, at the same time, discoverable and shareable.

For the exam, split it into two jobs. Data protection is the set of controls that keep data safe and compliant. Sharing and federation is how governed data reaches consumers and engines outside Snowflake.

Governed datapolicies, tags, lineageHorizon CatalogInside SnowflakeSQL, sharesandOutside SnowflakeIceberg REST, Marketplace

How it works

Establish and maintain data protection

Data protection is not one feature. It is a layered baseline that you set up once and then monitor.

LayerFeatures
Know your dataObject tagging, classification, lineage, data quality monitoring
Control accessRole based access control, masking, row access, projection and aggregation policies
Secure the accountNetwork policies, authentication and MFA, encryption of all data at rest and in transit
AuditACCESS_HISTORY, query history, login history
Detect driftTrust Center, which flags misconfigured roles and unprotected sensitive columns
RecoverTime Travel, Fail-safe, replication

Maintaining protection means repeating the loop.

  1. Classify new data and tag it.
  2. Attach policies, by tag where possible, so new columns are protected automatically.
  3. Review ACCESS_HISTORY for who read sensitive columns.
  4. Use the Trust Center and the governance dashboards in Snowsight to find gaps.
  5. Fix gaps, then repeat on a schedule.
Classifyfind sensitive datatagProtectpolicies by tagauditACCESS_HISTORYwho read whatdetectTrust Centergaps foundfixClassifyrepeat
Exam tip

Tag based policies are the scalable choice. Protecting 400 columns one by one is easy to get wrong, while one tag with one masking policy covers new columns automatically.

Share and federate data outside Snowflake

Horizon Catalog is built to hand governed data to others without copying it and without losing the rules.

NeedFeature
Share live data with another Snowflake accountSecure Data Sharing, a share of secure objects, no copy
Publish data products to many consumersSnowflake Marketplace listings and private listings
Share inside the companyInternal Marketplace, governed data products across teams
Let Spark, Flink or Trino read Snowflake tablesIceberg REST Catalog API on Snowflake managed Apache Iceberg tables
Read tables registered in another catalogCatalog-linked database to catalogs such as AWS Glue
Share across regions and cloudsListings with cross-cloud auto-fulfillment, or replication

Important ideas:

  • Open table format. Iceberg tables let other engines read and write the same data. Tables with masking and row access policies remain governed: Snowflake documents policy enforcement for Apache Spark reading through Horizon Catalog, so check what each external engine supports.
  • No copy for sharing. A consumer of a secure share reads the provider's live data, and the provider controls what is exposed through secure views, policies and the share contents.
  • Federation. A catalog-linked database lets Snowflake see tables that live in an external Iceberg catalog, without moving their files.
  • Governance travels. Tags and policies follow the data so that sharing does not create an unprotected copy.
Iceberg tableSnowflake managedHorizon REST APISpark or Trinoexternal enginepolicies enforcedGoverned readsame rules as SQL

Decision guide for a scenario:

ScenarioBest fit
A partner on Snowflake needs live accessSecure Data Sharing
Customers without Snowflake need dataReader account managed by the provider
The data science team uses SparkIceberg tables with the REST catalog
Data lives in Glue and you need it in SnowflakeCatalog-linked database
Many unknown consumersMarketplace listing
Watch out

Sharing exposes whatever the share contains. Use secure views or policies to limit columns and rows, and never share raw tables that include sensitive data.

Common mistakes

Treating governance as one feature. It is tags, policies, audit, monitoring and recovery working together.

Copying data to share it. Use a share, listing or Iceberg access, so there is one governed copy.

Protecting columns one by one. Use tag based policies for scale.

Assuming external engines bypass policies. Horizon Catalog offers policy enforcement for engines such as Spark; check support per engine rather than copying data out.

What to remember for the exam

  • Horizon Catalog covers discovery, governance, security and interoperability for data inside and outside Snowflake.
  • Protection baseline: classify and tag, policies, audit with ACCESS_HISTORY, monitor with the Trust Center.
  • Iceberg REST Catalog API opens Snowflake managed Iceberg tables to engines such as Spark, Flink and Trino for reads and writes; policy enforcement is documented for Spark.
  • Catalog-linked databases connect to external Iceberg catalogs; Marketplace and secure shares distribute data without copying.

Check yourself

  • How can Spark read a Snowflake managed Iceberg table and still honor masking rules?

    Through Horizon Catalog with Snowflake's documented Spark integration, which enforces the masking policies for that engine.

  • Which feature surfaces unprotected sensitive columns and risky roles?

    The Trust Center.

  • How do you give a partner live access to data without copying it?

    Use Secure Data Sharing, exposing only the secure objects you add to the share.

Topic 3 of 7 in Data Governance

Test yourself on Data Governance

The full Snowflake DEA-C02 question bank has 709 questions, with timed practice exams and an explanation for every answer.

This study guide is independent and is not affiliated with, endorsed by, or sponsored by Snowflake. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.