MAR 17, 2026|5 min read
Full bidirectional interoperability for Iceberg tables now in Snowflake Horizon Catalog
At Snowflake, our mission to mobilize the world’s data is built on a simple conviction: Data must be open, accessible and governed without forcing you to make trade-offs between flexibility and performance.
We have moved aggressively to make Apache Iceberg a first-class citizen in Snowflake. We introduced Snowflake managed Iceberg tables to give you the enterprise-grade performance of Snowflake’s engine on the storage you manage. We then broke down silos with catalog-linked databases, making it easier than ever to seamlessly read and write to tables managed by external catalogs. Together, customers can now access their entire Iceberg ecosystem from a single governed pane of glass.
Today, we are thrilled to announce the final piece of the puzzle: the ability to securely write to Snowflake managed Iceberg tables from external engines via Snowflake Horizon Catalog’s implementation of the open APIs, including Apache Polaris™ (in public preview).
This builds on external engine reads (generally available) and delivers full bidirectional interoperability for Iceberg tables in Snowflake.
Whether your data is managed by Snowflake or an external system, and whether you are using Snowflake’s engine or external engines such as Apache Spark or Trino, your data remains freely accessible, securely governed and truly open.
See the product documentation for details.
Why this matters
Many teams already standardize on Iceberg to enable multi-engine analytics and engineering. But in practice, “multi-engine” often turns into “multi-copies,” “multi-permissions” and “multi-audit trails.”
With external writes to Snowflake managed Iceberg tables, you can keep a single governed copy of Iceberg data in Snowflake — while letting external engines do what they do best for your use cases.
Interoperability in action: how Autofleet simplified operations and centralized governance with Horizon Catalog
Autofleet, a leading fleet management optimization platform for fleets and mobility operators, uses telematics data to launch customer-facing applications that enable usage-based vehicle fleet insights.
Using open data format (Iceberg) with Horizon Catalog, Autofleet operates at a scale ranging from 1,000 to 1.5 million vehicles while remaining cloud-agnostic — supporting both operational workloads and analytics.
Autofleet chose Snowflake and Horizon Catalog to simplify operations and centralize governance. Because Horizon is built on open APIs that now also support writes from external engines, Autofleet received full Iceberg interoperability (read and write) with external engines, while gaining access to enterprise controls and easier management.
The availability of external writes is a game changer. As part of our open lakehouse strategy at Autofleet — bringing together real-time processing and flexible data platforms — this solution fits perfectly with our roadmap. Its openness, scalability and enterprise-grade reliability make it a natural choice for our scenarios. We’re excited to move forward with it!
Pavel Millstein
Data Engineering Tech Lead at Autofleet
Open by design: Powered by Apache Polaris™
Our philosophy is "Open by Design" to give customers architectural choice. We believe you should stay with a vendor because you love the experience and it delivers value — not because you want to avoid costly migrations.
To achieve genuine openness, we didn't just build APIs that happen to support an open format. We integrated the goodness of Apache Polaris™ — the industry’s leading open source catalog for Iceberg — directly into the Snowflake Horizon Catalog so customers can rely on a catalog foundation aligned with open standards and a broader ecosystem without being locked in.
Because the integration is grounded in an open standard and open core, you get confidence that your architecture isn’t trapped behind proprietary interfaces. And you get the benefits of innovation by the community, not just one vendor. If you ever decide you want to run an open source Iceberg catalog yourself, Apache Polaris (now recognized as an Apache Top-Level Project) provides that option — without forcing you to redesign how your engines talk to the catalog. (We obviously hope you love the Horizon experience and never need to.)
This gives you ultimate peace of mind: You get the enterprise-grade scale, governance, security and compliance capabilities of Snowflake Horizon, backed by an open core that enables you to never face vendor lock-in for your data or metadata.
Completing the bidirectional interoperability
Interoperability is a two-way street. This launch opens lanes in both directions, providing a unified, flexible experience for all your Iceberg data:
| External engines writing to Snowflake (public preview) |
Snowflake writing to external catalogs |
|---|---|
Previously, external engines could read Snowflake managed tables. Now, they can write back to them.
|
Through catalog-linked databases, Snowflake can interact seamlessly with Iceberg tables where the source of truth is any external catalog.
|
Single-pane-of-glass governance with Snowflake Horizon
The true power of this interoperability isn't just moving data — it’s governing it. Snowflake Horizon serves as the single pane of glass for your entire data estate.
By centralizing on Horizon, you can apply consistent governance across all Iceberg tables — whether they are managed internally by Snowflake or linked externally. You gain unified access controls, automated discovery, data quality monitoring, sensitive data classification, tag-based masking policy, access history and lineage across your entire footprint, regardless of which engine is reading or writing the data.
Get started today
By supporting external writes, Snowflake gives you the ultimate flexibility to build your data architecture exactly how you want it, backed by the open core of Apache Polaris and the robust governance of Snowflake Horizon.
Ready to start writing to Iceberg tables from your favorite external engines?
- Review external engine reads (GA) and external engine writes (public preview): Documentation
- Learn about Horizon Catalog: Documentation
Join us in building a truly open data ecosystem. We can’t wait to see what you build.
Authors
Just For You
NOV 07, 2025|8 min read
JUN 09, 2025|6 min read
