Apache Polaris, the open source catalog for Apache Iceberg, officially graduated to a top-level Apache project this week. If you're running Iceberg tables (or planning to), this is a big deal.
Polaris has been incubating at the Apache Software Foundation since August 2024. In that time, the project shipped 6 releases, attracted around 100 contributors from across the industry, and closed over 2,800 pull requests. The graduation vote passed with unanimous support from binding voters, and the ASF Board has now approved it as a “top-level project.”
So what does "top-level project" actually mean? It confirms that Polaris has met the Apache Software Foundation's bar for community governance, technical maturity, and vendor neutrality. It's no longer an incubation experiment. It's a fully endorsed ASF project with its own Project Management Committee.
How Apache Polaris solves the Iceberg catalog gap
For a while, the Iceberg ecosystem had a catalog gap. You could store your tables in an open format, but the catalog layer that tracks metadata and enables multi-engine access? That was still mostly proprietary. AWS Glue, Unity Catalog, and others each came with their own trade-offs and lock-in risks.
Polaris fills that gap as a community-driven, open standard. It implements the Iceberg REST API spec, so any compliant engine can discover and query tables through a single catalog. That's the kind of interoperability that makes open lakehouse architectures actually work in practice, not just in conference talks.
As data architectures shift toward supporting both traditional analytics and AI workloads, having open and interoperable metadata infrastructure isn't a nice-to-have — it's table stakes.
Why Fivetran built our Iceberg REST Catalog on Apache Polaris
We built our Iceberg REST Catalog on Apache Polaris. It's the default catalog for every Iceberg table in Fivetran's Managed Data Lake Service, handling metadata across S3, ADLS, GCS, and OneLake destinations. When you query your Fivetran-managed Iceberg tables from Spark, Trino, Snowflake, or DuckDB, Polaris makes that work.
We integrated this project early because it implements the Iceberg REST catalog spec in a genuinely vendor-neutral way — with no lock-in to a single cloud provider or query engine. Your data stays open and accessible from whatever tools your team actually uses.
Graduation gives us even more confidence in that investment. A top-level Apache project comes with long-term governance guarantees, a diverse contributor base, and the kind of stability you want underneath production infrastructure.
What this means for Fivetran customers
Nothing changes about how you use the product today. Your Fivetran Iceberg REST Catalog already runs on Polaris, and it'll keep working exactly as it does now.
If anything, the graduation reinforces that the foundation underneath your data lake is built on mature, community-governed open source.
If you're not using Fivetran's Managed Data Lake Service yet, you can get started today. We handle the ingestion, schema evolution, compaction, and catalog management so you can focus on actually using your data.
[CTA_MODULE]
