Apache HOP is a free, open-source platform for building and running data pipelines. This guide explains what it is, who created it, what it’s used for, and how it’s licensed — written for people who are new to both HOP and Pentaho.
Every organisation moves data. From a transactional database into a warehouse, from an API into a report, from one cloud into another — usually with some cleaning, reshaping, and joining along the way. Doing this by hand, or with custom scripts, gets painful quickly. Apache HOP is one of the tools built to handle it at scale, and it has an interesting story behind it.
This article is a plain-language introduction: what HOP is, who made it, why it exists, what you can do with it, and how it’s licensed. No prior knowledge of HOP or Pentaho is assumed.
What is Apache HOP?
Apache HOP (the name comes from Hop Orchestration Platform) is a free, open-source platform for data integration and data orchestration. In practice, that means two things:
- Building data pipelines — the flows that read data from somewhere, transform it, and write it somewhere else.
- Orchestrating work — deciding when those pipelines run, in what order, and what happens if one fails.
You design pipelines visually, in a desktop application called HOP GUI, by dragging components onto a canvas and connecting them. Each component does one thing — read a file, filter rows, join two streams, write to a database — and you chain them together. The same pipeline can then run on your laptop during development and on a cluster in production without being rewritten.
HOP is a project of the Apache Software Foundation, the same organisation that governs Kafka, Spark, Airflow, and hundreds of other widely-used open-source projects. It is released under the Apache License, version 2.0.
The project’s own description is worth reading directly on the Apache HOP website.
Who created Apache HOP?
Apache HOP was started by Matt Casters, a Belgian software architect who has been working on data integration tools for more than two decades. If you have ever heard of Kettle or Pentaho Data Integration, Casters is the person who originally wrote them — he began developing Kettle around 2000–2002 and released it as open source in 2005.
He is currently a member of the Apache HOP Project Management Committee (PMC) and is recognised as one of the project’s founders.
Casters did not build HOP alone. A central role was played by know.bi, a Belgian data-engineering company whose team, in their own words, helped create Apache HOP “from its inception” and remain core contributors. know.bi describes its people as the platform’s own architects, and today offers commercial enterprise support for HOP under the product name Putki.
We, Toba Consulting, are know.bi’s official partner for the Asia-Pacific region, working alongside the team that architects HOP.
Why does Apache HOP exist?
To understand why HOP exists, a little history helps.
Kettle, the project Casters started in the early 2000s, was acquired and became the data-integration engine inside Pentaho, a commercial analytics company. Pentaho was in turn acquired by Hitachi in 2015. Over the years, Kettle grew into a large, mature codebase, but it also carried a lot of legacy design decisions and was tied to a single commercial vendor’s roadmap.
Around 2017–2019, the original engineering team — Casters together with the team at know.bi — set out to rebuild the core ideas from scratch, without the baggage. The goal was a platform that was modern, modular, and able to run anywhere, and crucially one that belonged to the community rather than to a single company.
That refactor was accepted into the Apache Incubator in September 2020 and graduated to become an Apache Top-Level Project in 2021. From that point, Apache HOP was governed by the Apache Software Foundation, meaning no one company can sell it, close it, or shut it down.
If you come from Pentaho: HOP is best understood as a spiritual and technical successor to Pentaho Data Integration, built by largely the same people. It is not a rebranded version of Pentaho — it is a separate codebase, written from scratch, that can import many existing Pentaho/Kettle jobs and transformations. (We cover the differences in more detail below.)
The main features
The features that tend to matter most when you are evaluating HOP:
- Visual, drag-and-drop design. You build pipelines and workflows in HOP GUI rather than writing boilerplate code. A pipeline moves and transforms data; a workflow controls when and how pipelines (and other tasks) run.
- Design once, run anywhere. A pipeline designed in HOP GUI can run on the native engine, or on Apache Spark, Apache Flink, Google Cloud Dataflow, or AWS EMR — HOP achieves this through Apache Beam. You don’t rewrite the pipeline when you change where it runs.
- Metadata-driven. In HOP, every object type describes how data is read, manipulated, or written. This makes it possible to generate pipelines dynamically from configuration rather than hand-building each one.
- Git-native, with no database repository. HOP stores projects as a folder of JSON files rather than inside a database. That means your pipelines live happily in Git alongside the rest of your code, with proper version history, branching, and code review.
- Built-in lifecycle management. You can manage development, test, and production environments from within HOP itself.
- A large library of transforms and actions for reading from and writing to databases, files, message queues, cloud storage, APIs, and more.
What is it used for?
HOP is general-purpose, so it shows up wherever data needs to move or be transformed. Common use cases include:
- ETL and ELT for data warehouses and data lakes — the classic “extract, transform, load” workload.
- Database migration and synchronisation between systems.
- Preparing data for reporting and dashboards.
- Automating repeatable business-data processes, such as scheduled financial calculations and regulatory reporting. For example, we have used HOP to automate PSAK 73 / IFRS 16 lease-accounting calculations for Telkom Indonesia.
- Cloud and big-data pipelines that need to run on Spark or Flink without being rewritten.
How Apache HOP differs from Pentaho
If you are completely new to both tools, you can safely skip this section. It is here for readers who keep hearing HOP and Pentaho mentioned in the same breath.
The short version: HOP and Pentaho Data Integration share an origin — the same original author and largely the same engineering team — but HOP is a separate, modern rewrite, not a new version of Pentaho.
A few concrete differences:
| Apache HOP | Pentaho Data Integration | |
|---|---|---|
| Origins | Built from scratch by the original Kettle team, ~2017 onward | Evolved from Kettle, matured inside Pentaho/Hitachi |
| Governance | Apache Software Foundation (vendor-neutral) | Commercial product, owned by a single company |
| How it stores work | Folder of JSON files (Git-native) | A database repository or files, depending on setup |
| Where it runs | Native, Spark, Flink, Dataflow, EMR (via Beam) | Primarily the JVM-based Pentaho engine |
| Design philosophy | Metadata-driven; pipelines generate from configuration | Transformation-and-job model from the Kettle era |
None of this means Pentaho is “bad.” It means the two tools were built in different eras, under different ownership models, for somewhat different expectations about how data engineering should work. Many teams run them side by side, and HOP can import a good portion of existing Pentaho .ktr and .kjb files.
Licensing, explained
Apache HOP is released under the Apache License, Version 2.0. For most readers, the practical implications are:
- Free to use, including in commercial environments and commercial products. There are no per-user, per-CPU, or per-server licence fees.
- Free to modify and redistribute, including as part of a proprietary product, as long as you follow the license’s attribution and notice requirements.
- No vendor lock-in at the license level. Because the project is governed by the Apache Software Foundation, the code belongs to the community. A company can be acquired or change strategy, and the software continues — the same guarantee that applies to Kafka, Spark, and Airflow.
It is worth separating two things that sometimes get confused. The open-source project (Apache HOP itself) is free. Commercial support, training, and enterprise tooling around HOP — such as know.bi’s Putki — are paid services offered by companies, but they are optional. You can run HOP in production without ever paying anyone.
Getting started
The fastest way to understand HOP is to try it. You can download HOP GUI, open it, and build a first pipeline in an afternoon — no server required. The Apache HOP website has the download, full documentation, and links to the community mailing lists.
Last reviewed: August 2026.
Sources & further reading
- Apache HOP — official site
- know.bi — core contributors to Apache HOP
- Matt Casters (Apache HOP PMC / founder) — LinkedIn
- “Introducing Apache Hop: Matt Casters gets the Kettle team back together” — The Stack
- “The Apache Software Foundation Announces Apache Hop as a Top-Level Project” — ASF news archive
- “Some Historic Cornerstones of Kettle & Pentaho” — kettle.bleuel.com