Blog
USE CASE
InteroperableLakehouse
AI is exposing the cost of fragmentation. Every copy adds pipelines, drifting policies and conflicting metrics. Build a fully managed, open and interoperable data platform tuned for AI and any Iceberg compatible engine in your stack.
Explore more interoperable lakehouse use cases:
LEADERS POWER THEIR LAKEHOUSE WITH SNOWFLAKE
Capture complete context
Define business meaning once, so every agent, analyst and tool works from the same definitions, portable through Apache Ossie (incubating).
Achieve universal governance
Govern every Apache Iceberg™ table where it lives, from one catalog to any compatible engine.
Access data in place
Cut costs with a single resilient, shareable and optimized copy that Snowflake maintains and tunes for every engine that reads it.
UNIVERSAL GOVERNANCE
Define governance in Snowflake. Extend it to every Iceberg table, on any catalog
Your teams have multiple catalogs and platforms, forcing you to reconcile governance policies. Horizon Catalog, built on Apache Polaris™, extends Snowflake RBAC to externally managed tables in place. Govern today, and migrate when a team is ready, if ever.
Govern tables in any catalog†
Access externally managed Iceberg tables from any engine through Horizon's IRC endpoints with Snowflake RBAC applied.
Protect sensitive data across engines†
Masking and row access policies apply to every query Snowflake serves or mediates through the Scan Plan API, regardless of engine.
Audit every engine interaction
Security teams review external-engine access across your data estate in one place with IRC telemetry in Event Tables and Audit History records each Horizon IRC interaction.
ONE DATA COPY
Run your engines on one optimized data copy
Cut the pipelines between platforms and engines. Read and write Snowflake-managed Iceberg tables from Snowflake or any engine, in Snowflake storage or your own bucket, and query Delta and Parquet where they live.
Tune layout for every engine
Set file size and partition layout for the engines that read each table, so Spark and Trino scan faster, not just Snowflake.
Automate table maintenance
Compaction, manifest compaction, clustering and orphan file cleanup† run automatically on Snowflake-managed tables, whether Snowflake, Spark or Flink wrote the data.
Stay online through outages
Replicate Snowflake-managed Iceberg tables across regions and clouds with failover groups, while client redirect keeps engines connected.¶
Hand off Iceberg storage
Create Iceberg tables any engine can read and write, with no buckets to manage. Tri-Secret Secure keeps your managed keys in your control on AWS and Azure.¶
Connect any catalog
Access any Iceberg table with read-write access, automatic discovery and refresh in a connected view of your data estate with catalog-linked databases.
Share without copies
With Open Data Sharing powered by Apache Polaris, consumers read your governed tables from any Iceberg REST client, no Snowflake account required.
Return to a known-good state
Tag snapshots such as month-end close, query a table as of a tag and test changes on isolated branches* that other Iceberg engines can read.
BUSINESS LOGIC
Give meaning to your datafor every agent and person
Ground AI and analytics in governed business definitions, wherever the data lives
Most enterprise data has no definition or, if they do, those definitions don’t agree, so dashboards and AI return numbers nobody trusts. Horizon Context gives every analyst, app and agent the same governed meaning, built from work your teams have done, and automates the rest, with you in control.
Learn from what you already have
Semantic View Autopilot and Cortex Sense learn business meaning from the queries, models and dashboards your teams already use, so Semantic Views start from what exists instead of a blank page.
Decide what's official†
Record approved definitions in Business Ontology, then compute them consistently with semantic views.
Carry meaning across your tool stack
Semantic views are portable through Apache Ossie (incubating), so compatible BI tools and AI systems work from one definition.
CUSTOMERS
Leading organizations build their open andinteroperable data foundation on Snowflake
Affirm achieves agency over its data and improves performance
Affirm turned to Snowflake for its financial workloads to improve operational efficiency and performance, while keeping the flexibility and choice of an open and interoperable architecture.
- 66% Faster SLAs for fresher data
- 6X Lower monthly replication pipeline costs
RESOURCES
Exploreinteroperable lakehouse resources
GET STARTED
Take the next stepwith Snowflake
Start your 30-day Snowflake trial today
- $400 in free credits
- Immediate access to the AI Data Cloud
- Enable your most critical data workloads
