# Apache Hudi

> A table format built for fast updates. Jobs can read just what changed.

**Field:** Data Engineering  
**Category:** Storage & Warehousing  
**Also known as:** Hudi, Hadoop Upserts Deletes and Incrementals

## Think of it like

It is like a notebook where you fix one line. You do not copy out the whole page.

## Definition

Apache Hudi is an open table format built for frequent updates in a data lake. It was first built at Uber. It is good at upserts, which update a row if it exists or add it if not. It also lets a job read only the rows that changed since its last run. It offers two table types, Copy-on-Write and Merge-on-Read.

## Example

A ride app updates millions of trip rows as fares are settled. Hudi applies the changes, and the next job reads only the changed trips.

## Why it matters

Many lakes get a stream of changes, not new rows only. Hudi is shaped around that case.

## Common mistake

Picking a table type at random. Merge-on-Read writes fast but reads slower. Copy-on-Write is the other way round.

## Three principles

- **Upserts.** Update if there, add if not.
- **Incremental reads.** Read only what changed.
- **Two table types.** Copy-on-Write or Merge-on-Read.

## Related terms

- [Open Table Format](https://learn.benibix.com/term/data/open-table-format)
- [Apache Iceberg](https://learn.benibix.com/term/data/apache-iceberg)
- [Delta Lake](https://learn.benibix.com/term/data/delta-lake)
- [Data Lake](https://learn.benibix.com/term/data/data-lake)

---

Source: https://learn.benibix.com/term/data/apache-hudi
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-16
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
