Apache Hudi V2

Apache Hudi 1.0 stable October 2024.

Definition

Hudi 1.0 stabilizes table format spec, adds optimized metadata table, secondary indexes (with Apache Lucene-based record index), expression index, precise time travel queries. Two table types: Copy-on-Write (CoW) and Merge-on-Read (MoR). Compatible with Apache Spark, Apache Flink, Apache Hive, Trino, Presto.

Origin

Uber Engineering 2016 (Vinoth Chandar), donation to Apache Software Foundation 2019, Top-Level Project May 2020.

Usage

Adopted by Uber, Robinhood, Walmart Labs, ByteDance. Strength in streaming upserts + CDC (Change Data Capture) use cases.

Related terms

• Apache Hudi — fiche socle.
• Apache Iceberg V3 — concurrent.
• Delta Lake V4 — concurrent.