Skip to main content

Best data catalogs for Snowflake

Emily Winks, Data Governance Expert, Atlan
Data Governance Expert
Updated:
|
Published:
22 min read

Key takeaways

  • Horizon Context, with metadata connectors to 18 sources, has been in Preview since Oct 12, 2026.
  • Snowflake external lineage through OpenLineage became generally available on Sep 3, 2026.
  • Alation, Atlan, Collibra, DataHub and OpenMetadata each document Snowflake coverage in their own terms.
  • Snowflake native lineage and Atlan tag sync both rely on Enterprise Edition or higher.

What are the best data catalogs for Snowflake?

Snowflake Horizon Catalog is the native baseline in every Snowflake account, with features that depend on your edition. Its Horizon Context feature, in Preview since Oct 12, 2026, adds metadata connectors to 18 sources. Alation, Atlan, Collibra, DataHub and OpenMetadata each document Snowflake integrations. Pick by your Snowflake edition, the sources you run outside Snowflake, whether you want to operate the catalog yourself, and whether AI agents will query it.

The six catalogs compared:

  • Snowflake Horizon Catalog: the native baseline, with Horizon Context in Preview since Oct 12, 2026
  • Alation: Snowflake OCF connector with table-level lineage from ingested query history
  • Atlan: AI Context Platform with column-level lineage from Snowflake query history, tag import and reverse sync to existing Snowflake tags
  • Collibra: technical lineage for Snowflake and outbound tag mapping in public preview
  • DataHub: open-source Snowflake source with table-level and column-level lineage
  • OpenMetadata: Snowflake connector listing lineage, data quality and auto-classification

Not sure how ready your catalog stack is?


Snowflake Horizon Catalog is the native baseline for every Snowflake account, and five third-party catalogs document Snowflake integrations: Alation, Atlan, Collibra, DataHub and OpenMetadata. Snowflake describes Horizon Catalog as covering data “inside or outside of Snowflake,” and its Horizon Context feature has been in Preview since Oct 12, 2026, with metadata connectors to 18 sources.

Map Your Snowflake Catalog Coverage


Give it every system in your estate and which catalog governs each today. It returns a coverage table and the specific gap ranked by what breaks first. Read the skill.

Paste into a new chat

Use the skill at https://atlan.com/skills/catalog-coverage-gap-map.md to map catalog coverage across our Snowflake estate and the platforms around it. Ask me for whatever it needs.

Run once in a terminal

curl -fsSL --create-dirs \
  -o ~/.agents/skills/catalog-coverage-gap-map/SKILL.md \
  https://atlan.com/skills/catalog-coverage-gap-map.md

For an agent

curl -fsSL https://atlan.com/skills/catalog-coverage-gap-map.md

If you run a Snowflake platform or data governance team, the catalog you pick decides whether lineage, tags and definitions reach the people and AI agents who query your data, or whether they work from wrong or missing context. Start with Horizon Catalog if your estate lives mostly inside Snowflake, and add a third-party catalog when your sources sit outside its 18 Horizon Context connector sources or you want one catalog across several warehouses and BI tools. Horizon comes first below as the native baseline, and the five third-party catalogs follow in alphabetical order, unscored, because each vendor’s docs describe Snowflake coverage in different terms.

Every status below comes from a dated release note. Undated docs pages were checked in November 2026.

Dated release-note statuses, checked November 2026

Catalog What its docs describe for Snowflake Status as stated in a dated release note
Snowflake Horizon Catalog Horizon Context, with metadata connectors to 18 sources Preview, Oct 12, 2026
Snowflake Horizon Catalog External lineage through the OpenLineage standard Generally available, Sep 3, 2026
Snowflake Horizon Catalog Snowflake-provided tags for cost center, certification status, sensitivity, environment and project Generally available, Oct 20, 2026
Collibra Outbound mapping of Collibra tags, classifications and other fields to Snowflake tags Public preview, Oct 18, 2026 release-note block

What counts as a data catalog for Snowflake?

A data catalog for Snowflake is a product that reads Snowflake metadata so people can find, describe, tag and trace your objects, and Snowflake ships one of its own. Third-party catalogs read from the same account and add their own layer on top.

The native one is Horizon Catalog. Snowflake calls it “the agentic catalog for all your data” and says it is “compatible with any engine, any data, and any cloud.”

Inside it sits Horizon Context, which Snowflake defines (Snowflake, 2026) as “the governed context layer built into Horizon Catalog.” It works in four layers: structural, operational, semantic and behavioral.

Tags are one surface all of these products read. Per Snowflake’s tagging docs (Snowflake, 2026), a tag is a schema-level object you assign to another Snowflake object, with its value stored as a string key-value pair. One object can carry many tags.

Edition matters before any comparison. Snowflake lineage needs Enterprise Edition or higher, and so does Access History (Snowflake, 2026), the ACCESS_HISTORY view that Atlan’s lineage, Collibra’s SQL-API mode and OpenMetadata’s default lineage path all read. Creating and setting tags works on every edition, per Snowflake’s tagging docs, but Atlan’s tag sync reads Snowflake’s ACCOUNT_USAGE tag views, which are available on Enterprise Edition and later.

For how a catalog relates to the wider context layer, see data catalog vs context layer, what a context catalog is, and how a context layer for Snowflake sits above the warehouse. Whichever term you use, the comparison below starts from what each product’s own documentation says it reads from your account.


Best data catalogs for Snowflake: the comparison table

Snowflake Horizon Catalog comes first as the native baseline, and Alation, Atlan, Collibra, DataHub and OpenMetadata follow in alphabetical order. We did not score them, because each vendor’s own docs describe Snowflake coverage in different terms and a score would rest on cells that do not compare like with like.

The six catalogs side by side


Catalog Snowflake objects its docs list Lineage source its docs name Tags Also in its docs
Snowflake Horizon Catalog (native baseline) Snowflake objects and Iceberg tables via the Iceberg REST Catalog API Native lineage of data movement and dependencies; OpenLineage events for external lineage Object tags as schema-level key-value objects Metadata connectors to 18 sources (Preview, dated above)
Alation Tables, columns, views, external tables, procedures, functions, dynamic tables and data shares Ingested query history at table level; column level through a parser add-on Extracts Snowflake tags Row access and masking policies with Policy Center enabled; a core connector on all Alation tiers
Atlan Databases, schemas, tables, views, procedures, stages, streams, pipes, UDFs, dynamic tables and Iceberg properties Query history (QUERY_HISTORY and ACCESS_HISTORY), at column level Imports object tags from Horizon; reverse sync updates existing Snowflake tags Atlan Lakehouse, read by Snowflake through a catalog-linked database; hosted Atlan MCP
Collibra A schema-level view in SQL mode; Snowflake semantic views in Data Lineage Executed queries through ACCESS_HISTORY in SQL-API mode Outbound tag mapping, public preview (Oct 18, 2026) Description mapping to the Snowflake COMMENT field, generally available (Oct 18, 2026)
DataHub Schema metadata, descriptions, domains, usage and optional profiling View definitions, dynamic table definitions and SQL query history, at table and column level Optional, through extract_tags A semantic_views setting in its Snowflake source
OpenMetadata Metadata, stored procedures, owners, sample data and query usage ACCESS_HISTORY by default, at table and column level, with query-log parsing as a fallback From account_usage tag_references Data quality, profiler, auto-classification and dbt

How to read the table


Docs pages for Alation, Atlan, DataHub and OpenMetadata are undated. Checked November 2026.

Read each row against your own estate, not against the row above it. If you are building a formal evaluation, the data catalog RFP template turns these columns into vendor questions, and context layer evaluation criteria adds the agent-facing tests. The Microsoft Fabric catalog comparison follows the same native-first pattern for a Fabric estate.


Snowflake Horizon Catalog: the native baseline

Snowflake Horizon Catalog ships with every Snowflake account, though what it offers depends on your edition, and Horizon Context is its newer cross-platform layer. A few parts carry a dated status, and the rest are described in undated docs.

What Horizon Catalog and Horizon Context cover


Per Snowflake’s Horizon overview (Snowflake, 2026), Horizon Catalog lists “sensitive data protection, data quality, end-to-end lineage, AI guardrails, and AI governance.” Snowflake adds that “governance policies execute at the query engine layer, not the application layer.”

The same page covers Apache Iceberg. Snowflake says its Iceberg tables are readable and writable from any engine, and names Spark, Flink and Trino as engines that use its Iceberg REST Catalog API; test each engine you plan to use.

Catalog-linked databases connect to AWS Glue and Azure OneLake. For background on the format, see Apache Iceberg v3 for Snowflake teams.

Horizon Context is the newest piece. According to Snowflake’s Oct 12, 2026 release note (Snowflake, 2026), Horizon Context is in Preview, with metadata connectors to 18 sources, including Power BI, Tableau, dbt, Databricks, BigQuery, SQL Server, Oracle and PostgreSQL. The note says you can browse and search external assets, trace column-level lineage across platforms, and manage tags and descriptions in Horizon Catalog.

Snowflake’s connectors page (Snowflake, 2026) labels the connectors a Preview Feature, available to all accounts. Snowflake’s docs say there is no charge for adding connectors, ingesting metadata or storing it. Two costs still apply, per the same page: your active warehouse for some Snowsight steps, and compute on the external source when connectors query it. Each connected source refreshes every 24 hours.

How it lines up with a third-party layer is covered in Horizon Context and the Atlan context layer, and Snowflake Summit 2026 announcements gives the wider product picture.

Lineage, tags and classification


Snowflake’s native lineage tracks how data flows from source to target objects. It captures data movement, such as CTAS, INSERT and MERGE, and object dependencies, on Enterprise Edition or higher.

External lineage reached general availability on Sep 3, 2026, per Snowflake’s release note (Snowflake, 2026). It uses OpenLineage: tools such as dbt and Apache Airflow send OpenLineage events to a Snowflake REST endpoint. Snowflake’s Horizon Context docs add that any OpenLineage producer can post lineage events there.

Two tagging and classification updates carry dates:

Snowflake Open Catalog


Snowflake Open Catalog (Snowflake, 2026) is “a catalog implementation for Apache Iceberg tables,” built on the Apache Iceberg REST protocol, and “a managed service for Apache Polaris.” Snowflake’s docs add a sign-up rule: only customers who already have an Open Catalog account can keep using it and create more accounts, and “New customers should use Snowflake Horizon Catalog for Apache Iceberg tables and multi-engine interoperability with Iceberg.”

On the Horizon side, access to externally managed Iceberg tables through the Horizon Iceberg REST Catalog API became generally available on Oct 29, 2026, per Snowflake’s release note. Spark, Trino, DuckDB and PyIceberg can read and write those tables, and Snowflake RBAC applies. Access to a table with a Snowflake data protection policy remains in preview.

For how the Iceberg catalogs differ, see Apache Polaris vs Unity Catalog vs Glue vs Snowflake Open Catalog.

Semantic views


A semantic view (Snowflake, 2026) is a schema-level object that defines business metrics and models business entities and their relationships. Snowflake’s Horizon page says you can create them from scratch with Autopilot or ingest them from Tableau and Power BI. Separately, semantic view materializations became generally available on Nov 11, 2026.

The Snowflake semantic views guide covers the object in depth, and Snowflake semantic views vs Cube vs AtScale compares it with standalone layers. Where business definitions should live is its own decision: see warehouse-native vs universal semantic layers and the Open Semantic Interchange effort.

For an estate that lives mostly inside Snowflake, Horizon Catalog is the place to start. The third-party question opens when the sources you run sit outside the 18 Horizon Context connector sources, or when you want one catalog across several warehouses and BI tools.


Alation: Snowflake through the OCF connector

Alation documents its Snowflake integration as the Snowflake OCF Connector, part of its Open Connector Framework. The connector pages, checked in November 2026, are undated.

What Alation’s Snowflake docs list:

  • A core connector. Alation’s docs say “Core connectors are included with all Alation platform tiers (subject to each tier’s connector limits) and are fully supported by Alation.”
  • Coverage for both Alation Cloud Service and customer-managed instances of Alation.
  • Extracted objects that include schemas, tables, columns, views, external tables, stored procedures, functions and dynamic tables, plus Snowflake tags.
  • Table-level lineage calculated from query history ingested through query log ingestion (QLI), metadata extraction (MDE) and Compose queries.
  • Column-level lineage, which the lineage page says uses the Snowflake column-level lineage parser add-on.
  • A “Snowflake Data Shares Extraction” page.

Good to know:

  • Row access and masking policy extraction is listed in the connector scope, with policy information available when Policy Center is enabled in the Governance App.
  • Alation says its Cloud Service documentation moved to a new documentation site on Aug 18, 2026. The pages reviewed here sit on the site that continues to host customer-managed documentation, so Cloud Service buyers should check the new site too.

Alation’s wider platform direction is covered in what Alation’s AIOS is, and Alation vs Informatica compares Alation with Informatica.


Atlan: an AI Context Platform reading Snowflake metadata

Atlan is an AI Context Platform: it crawls Snowflake metadata, builds column-level lineage, imports and reverse-syncs tags, and serves that context to AI clients through Atlan MCP. Its Snowflake docs carry no page date; we read them in November 2026.

What Atlan’s Snowflake docs list:

  • Per Atlan’s crawl reference (Atlan, 2026), “Atlan crawls Snowflake metadata for databases, schemas, tables, views, and procedures, mapping each property to its Atlan asset and building column-level lineage.” The same page maps stages, streams, pipes, UDFs, dynamic tables and Iceberg table properties.
  • Column-level lineage built automatically from query history, reading QUERY_HISTORY and ACCESS_HISTORY, per how Snowflake metadata extraction works.
  • Object tags imported from Horizon, matched by tag name with case ignored, with an optional reverse sync that pushes tag updates from Atlan back to Snowflake. Per Atlan’s tag guide, reverse sync “only updates existing tags in Snowflake. It neither creates nor deletes any tags in Snowflake.”
  • Data quality monitoring that uses Snowflake’s native data metric functions.

Good to know:

  • The Atlan Lakehouse is built on Apache Iceberg with Apache Polaris as its Iceberg REST catalog. Snowflake reads it through a catalog-linked database; Atlan’s docs say “You need to bring your own Iceberg REST-compatible compute.”
  • Atlan MCP (Atlan, 2026) is a hosted server that lets AI clients such as Claude, Cursor and ChatGPT use Atlan as a context layer through the Model Context Protocol. See the MCP-connected data catalog and MCP for data lineage.
  • Tag sync relies on ACCOUNT_USAGE.TAGS and ACCOUNT_USAGE.TAG_REFERENCES, and lineage reads ACCESS_HISTORY. Both sit on Snowflake Enterprise Edition or higher.

Collibra: lineage and tag mapping for Snowflake

Collibra documents Snowflake through Collibra Data Lineage and a Snowflake outbound integration. Its technical lineage page is dated Jun 17, 2026, and its release notes carry dated entries.

What Collibra’s Snowflake docs list:

  • Two ingestion methods for technical lineage. SQL mode gives “a schema-level view.” SQL-API mode builds lineage “based on queries that were actually executed in Snowflake during the selected time period,” and relies on ACCESS_HISTORY.
  • Snowflake semantic views in Data Lineage. Collibra’s release notes say on Apr 7, 2026 that “Collibra Data Lineage now supports Snowflake Semantic Views,” and on Apr 19, 2026 that it “harvests Snowflake Semantic Views directly from the database.” Both entries are listed as generally available.
  • A Snowflake outbound integration. In the Oct 18, 2026 release-note block, Collibra says you can map Collibra asset status, articulation score, average rating, Collibra tags, responsibilities and data classifications to Snowflake tags, in public preview.

Good to know:

  • Description mapping to the Snowflake COMMENT field is generally available, per the same Oct 18, 2026 block.
  • SQL-API mode reads executed queries over a time period you select, so the period you pick shapes the lineage you see.
  • Collibra’s release notes from May 6, 2026 list tools added to the Collibra MCP servers, so Collibra also documents an MCP route for AI clients.

DataHub: open-source ingestion with a Snowflake source

DataHub’s code is published under the Apache 2.0 license on GitHub, and its docs include a Snowflake ingestion source. The source page, checked in November 2026, is undated.

What DataHub’s Snowflake docs list:

  • Column-level lineage, “Enabled by default, can be disabled via configuration include_column_lineage.”
  • Table-level lineage.
  • Data profiling, optional.
  • Dataset usage.
  • Descriptions and schema metadata.
  • Tag extraction, optional through extract_tags.
  • Domains.

Good to know:

  • DataHub’s Snowflake source documents a semantic_views setting, with its own options for usage and queries. Read the configuration block before you plan semantic view coverage.
  • DataHub’s docs say lineage edges are extracted from view definitions, dynamic table definitions and SQL query history, and lineage and tags are configuration choices in the ingestion recipe.
  • If you run the open-source edition yourself, your team operates the service. Open source vs commercial data catalog total cost of ownership covers that trade.

For the wider field around this project, see DataHub alternatives.


OpenMetadata: open-source connector with lineage and classification

OpenMetadata’s Snowflake connector page lists twelve features, and the project’s code is published under the Apache 2.0 license on GitHub. Like DataHub’s source page, the connector page carries no date; we read it in November 2026.

What OpenMetadata’s Snowflake docs list:

  • Metadata and query usage.
  • Data profiler and data quality.
  • Lineage and column-level lineage.
  • dbt.
  • Stored procedures and owners.
  • Tags, ingested from snowflake.account_usage.tag_references.
  • Sample data and auto-classification.

Good to know:

  • Tags come from Snowflake’s account_usage views, so check which Snowflake edition and role your connector uses.
  • For lineage, OpenMetadata’s docs say it uses Snowflake’s ACCOUNT_USAGE.ACCESS_HISTORY view by default and falls back to parsing query logs if the configured role cannot read that view.
  • As with DataHub, running the open-source edition yourself means your team runs the service, which changes the total cost comparison with a commercial product.

How does lineage reach each catalog for a Snowflake estate?

Each catalog’s docs name their own source for Snowflake lineage, and the source shapes what the lineage can show. Alation, Atlan, Collibra’s SQL-API mode, DataHub and OpenMetadata all work from Snowflake query history or ACCESS_HISTORY, and Snowflake itself accepts OpenLineage events from outside tools.

Catalog Lineage source its docs name Date or requirement in the docs
Snowflake Horizon Catalog Native lineage from data movement and object dependencies; OpenLineage events for external lineage Enterprise Edition or higher; external lineage generally available Sep 3, 2026
Alation Query history ingested through QLI, MDE and Compose; column level through a parser add-on Docs undated, checked November 2026
Atlan Query history, reading QUERY_HISTORY and ACCESS_HISTORY, at column level Docs undated, checked November 2026
Collibra Executed queries through ACCESS_HISTORY in SQL-API mode; a schema-level view in SQL mode Page dated Jun 17, 2026
DataHub View definitions, dynamic table definitions and SQL query history; column level on by default and configurable Docs undated, checked November 2026
OpenMetadata ACCESS_HISTORY by default, with query-log parsing as a fallback Docs undated, checked November 2026

Lineage built from executed queries shows what ran in the period the catalog read. Lineage from OpenLineage events shows what your orchestration tools report. A catalog can combine both, so ask each vendor which inputs feed each lineage edge and whether you can see where an edge came from.

Automated SQL lineage vs manual lineage mapping explains the query-parsing side, Airflow and OpenLineage explains the event side, and data lineage for AI covers why an AI answer needs either one.


What does a Snowflake catalog cost?

The docs reviewed here describe cost in different units, so collect a quote for your own estate before you compare. What the docs do say:

  • Snowflake’s Horizon Catalog connectors page says there is no charge for adding connectors, ingesting metadata or storing it. Your active warehouse still runs some Snowsight steps, and the external source bills for the compute that connector queries use.
  • Alation’s Snowflake OCF connector is a core connector, which Alation says is included with all platform tiers, subject to each tier’s connector limits.
  • DataHub and OpenMetadata publish their code under the Apache 2.0 license. If you host either yourself, the cost is your infrastructure and the people who run it.
  • For Atlan and Collibra, ask each vendor for a quote based on your Snowflake estate and the sources around it.

Snowflake compute is part of every option. Atlan’s lineage, Collibra’s SQL-API mode and OpenMetadata’s default lineage path all read Snowflake’s ACCESS_HISTORY view, and catalogs that profile or sample data run queries in your account. The open source vs commercial data catalog total cost of ownership guide covers the wider comparison.


How do you choose a catalog for a Snowflake estate?

Four questions narrow a Snowflake catalog shortlist quickly: which edition you run, what lives outside Snowflake, who will operate the catalog, and whether AI agents will query it. None of them names a winner; each points you to the docs that settle it for your estate.

Question What to check Where the docs say it
Which Snowflake edition do you run? Native lineage and ACCESS_HISTORY need Enterprise Edition or higher. Atlan’s tag sync reads ACCOUNT_USAGE tag views, available on Enterprise Edition and later Snowflake lineage and Access History docs; Atlan’s Snowflake tag docs
What runs outside Snowflake? List every source and BI tool, then compare it with the 18 Horizon Context connector sources (Preview) and each vendor’s own source list Snowflake’s connectors page; each vendor’s connector docs
Who will operate the catalog? DataHub and OpenMetadata publish their code under Apache 2.0, so your team can host it, and the license alone does not tell you which managed options exist. Alation documents Alation Cloud Service and customer-managed instances. Ask every vendor which deployment and support options apply GitHub repositories; vendor docs
Will AI agents query the catalog? How context reaches an agent: Snowflake’s Horizon Context inside Snowflake, or a protocol such as MCP, which Atlan MCP and the Collibra MCP servers use. Ask each vendor which agent route its docs describe Horizon Context docs; Atlan MCP docs; Collibra release notes

Further reading for the edition, estate and agent questions


The second question carries the most weight for a multi-platform estate. If your definitions and lineage need to reach beyond the warehouse, read how to extend your ontology beyond Snowflake and whether a data catalog can sit on top of existing catalogs without replacing Horizon. Context layer vs data catalog vs semantic layer separates the three jobs.

For the agent question, start with how to prepare Snowflake data for AI; the Atlan and Snowflake architecture page shows where a third-party layer runs beside the warehouse. Cortex Analyst vs custom text-to-SQL shows how answer quality depends on the context behind a query.

If you already own a catalog, extending it for AI readiness may beat a replacement, and the data governance platform evaluation criteria give you tests to run in a proof of value.

Test the shortlist on your own account


Run three tests before you sign anything:

  • Pick one schema that has dbt models and a BI dashboard on top. Write down the edges you expect, including one MERGE and one dynamic table, then count how many each catalog draws.
  • Tag one test object in Snowflake, record how long the tag takes to appear in each catalog, then change the value in the catalog and check what reaches Snowflake.
  • Query the catalog as a restricted role, and as an AI client if agents are in scope, and record what each one returns that the role should not see.

The catalog coverage gap map takes your list of systems, maps each one to the catalog that governs it, and names the gap that breaks first.


FAQs about data catalogs for Snowflake

Eight questions buyers ask about Snowflake catalogs, answered from Snowflake, Atlan and vendor docs.

1. Does Snowflake have a data catalog?


Yes. Snowflake Horizon Catalog is built into Snowflake, and Snowflake describes it as a catalog for data inside or outside of Snowflake. Horizon Context, part of Horizon Catalog, collects metadata across your data estate and organizes it in four layers: structural, operational, semantic and behavioral. Third-party catalogs read from Snowflake alongside it.

2. What is the difference between Snowflake Horizon Catalog and Open Catalog?


Horizon Catalog is Snowflake’s built-in catalog for governance, discovery and Iceberg interoperability. Snowflake Open Catalog is a managed service for Apache Polaris, built on the Apache Iceberg REST protocol. Snowflake’s docs say only customers who already have an Open Catalog account can create more, and point new customers to Horizon Catalog for Iceberg tables.

3. Can Snowflake’s catalog cover sources outside Snowflake?


Yes, through more than one route. Snowflake’s Oct 12, 2026 release note introduced Horizon Context in Preview, with metadata connectors to 18 sources, including Power BI, Tableau, dbt, Databricks and BigQuery, refreshed every 24 hours. Separately, external lineage through OpenLineage has been generally available since Sep 3, 2026.

4. Do Snowflake lineage and tags need Enterprise Edition?


Lineage, yes. Snowflake’s docs list data lineage and Access History as Enterprise Edition features. For tags, Snowflake says creating and setting tags is available to all accounts, while some advanced tag capabilities need Enterprise Edition or higher. Atlan’s Snowflake tag sync reads the ACCOUNT_USAGE.TAGS and ACCOUNT_USAGE.TAG_REFERENCES views, which Atlan’s docs say are available in Enterprise Edition and later. Check your edition before you compare lineage or tag features.

5. Is Snowflake external lineage generally available?


Yes. Snowflake’s release note of Sep 3, 2026 says external lineage is generally available and no longer in Preview. It uses the OpenLineage standard: tools such as dbt and Apache Airflow send OpenLineage events to a Snowflake REST endpoint, and column lineage arrives through the columnLineage facet.

6. How does Atlan sync Snowflake tags?


Atlan imports object tags from Snowflake Horizon. It matches Snowflake tag names to Atlan tags, ignoring case, and applies the tag and its value to the matching assets. If you turn on reverse sync, tag updates made in Atlan are pushed back to Snowflake; Atlan’s docs say reverse sync only updates existing tags in Snowflake and neither creates nor deletes tags.

7. How does Snowflake read the Atlan Lakehouse?


Through a catalog-linked database. The Atlan Lakehouse is built on Apache Iceberg and uses Apache Polaris as its Iceberg REST catalog. Atlan’s docs describe a one-time catalog integration that an administrator creates in Snowflake, after which Snowflake roles can query Atlan metadata in standard SQL. Atlan’s docs say you bring your own Iceberg REST-compatible compute, such as Snowflake.

8. Which open-source catalogs have a Snowflake connector?


DataHub and OpenMetadata both document one, and both projects publish their code under the Apache 2.0 license on GitHub. DataHub’s Snowflake source lists table-level and column-level lineage, usage, profiling and optional tag extraction. OpenMetadata’s connector lists metadata, query usage, lineage, column-level lineage, data quality, tags and auto-classification, and reads lineage from Snowflake’s ACCESS_HISTORY view by default.


Sources

  1. Snowflake Horizon Catalog, Snowflake documentation
  2. Horizon Context, Snowflake documentation
  3. Horizon Context (Preview), Snowflake release notes, Oct 12, 2026
  4. Horizon Catalog connectors, Snowflake documentation
  5. External lineage (General availability), Snowflake release notes, Sep 3, 2026
  6. Data lineage, Snowflake documentation
  7. Access History, Snowflake documentation
  8. Introduction to object tagging, Snowflake documentation
  9. Snowflake-provided tags (General availability), Snowflake release notes, Oct 20, 2026
  10. Autopilot for sensitive data classification (Public Preview), Snowflake release notes, Oct 26, 2026
  11. Snowflake Open Catalog overview, Snowflake documentation
  12. Externally managed Iceberg tables through the Horizon Iceberg REST Catalog API (General availability), Snowflake release notes, Oct 29, 2026
  13. Semantic views overview, Snowflake documentation
  14. Semantic view materializations (General availability), Snowflake release notes, Nov 11, 2026
  15. What does Atlan crawl from Snowflake, Atlan documentation
  16. How Snowflake metadata extraction works, Atlan documentation
  17. Manage Snowflake tags, Atlan documentation
  18. Snowflake data quality, Atlan documentation
  19. What is Atlan Lakehouse, Atlan documentation
  20. Connect Snowflake to Lakehouse, Atlan documentation
  21. Atlan MCP overview, Atlan documentation
  22. Snowflake OCF Connector, Overview and Enable Lineage pages, Alation documentation (docs.alation.com), accessed November 18, 2026
  23. Technical lineage for Snowflake ingestion methods, Collibra documentation (productresources.collibra.com), June 17, 2026
  24. Collibra release notes archive (Release 2026.04 page), entries dated Apr 7, Apr 19, May 6 and Oct 18, 2026, Collibra documentation (productresources.collibra.com)
  25. Snowflake ingestion source, DataHub documentation (docs.datahub.com), accessed November 17, 2026
  26. Snowflake connector, OpenMetadata documentation (docs.open-metadata.org), accessed November 17, 2026
  27. datahub-project/datahub repository, GitHub (github.com), accessed November 17, 2026
  28. open-metadata/OpenMetadata repository, GitHub (github.com), accessed November 17, 2026

Share this article

signoff-panel-logo

Atlan is the Context Layer for AI. It translates business knowledge, including data definitions, working procedures, and governance policies, into context AI can actually use. This knowledge lives in a single Enterprise Data Graph that every team and AI agent can reach.

In Atlan's research report, rich semantic metadata improved the accuracy of AI-generated SQL by 38%.

Atlan is recognized as a Leader across multiple Gartner reports and Forrester Waves, and is deployed at enterprises including Mastercard, Workday, General Motors, CME Group, HubSpot, FOX, Virgin Media O2, and Elastic.

Bridge the context gap.
Ship AI that works.