Key Features of Apache Paimon 0.9.0 Release

Apache Paimon version 0.9.0 has been available for a week, and this article highlights the most important new features of the release.

The core changes announced by the Paimon team include:

Paimon Branch – the Branch feature is now production‑ready, with a new scan.fallback-branch option that simplifies unified stream‑batch storage.

Universal Format – native Iceberg compatibility is added; enabling Iceberg mode generates compatible snapshots that can be read by the Iceberg ecosystem.

Caching Catalog – table metadata and manifest files are cached in the catalog by default, accelerating OLAP query performance.

Bucketed Append – the small‑file problem is largely mitigated, and the format can be used for Bucketed Joins in Spark, reducing shuffle.

Append Table DML – DELETE, UPDATE, and MERGE‑INTO operations are now supported for Append tables via Spark SQL, with optional Deletion Vectors mode.

Two features deserve special attention:

Paimon Branch

Previously Paimon offered tag support; now it provides a Git‑like branch capability, allowing multiple branches and data replacement between them, which can be used for data correction and historical snapshot retention.

Bucketed Append Table

Bucketed Append tables differ from regular Append tables by using a bucket key to hash data, ensuring ordered data within each bucket and enabling efficient Bucketed Joins. The new version prevents creation of bucketed tables without a bucket key, encouraging users to avoid this configuration unless necessary.

Overall, version 0.9 focuses on usability, and users are encouraged to consult the official documentation for detailed guidance.