Best way to deploy an Hop project to a customer?

[nazzo99]

Hi everyone,
I’m new to the world of ETL, and this is my first time deploying an ETL project using Apache Hop; so, I apologize in advance if my questions seems silly.

I’ve built a complete Apache Hop project for my company, and now I need to figure out and test how to deploy it to a client.

• How is a Hop project typically deployed to a client, and what, in your experience, are the best approaches?
• Is there a way to protect the project logic and scripts (SQL, JavaScript) once they’ve been deployed to the client?

Thanks in advance!

[AbdullahTammar]

For a first deployment that runs scheduled batches, I would start with Hop Run on a dedicated client machine, using the same Hop version and required plugins/JDBC drivers as your tested environment.

• Deploy the project files, metadata and supporting scripts together.
• Put client-specific connection values and paths in a separate environment configuration, with credentials supplied securely on the target machine.
• Register the project/environment there, then run the entry workflow through Hop Run. Your scheduler can capture its logs and exit status. For example, on Linux, after configuring the names and paths below:

./hop-run.sh -j client-project -e production -r local -f /opt/client-project/main.hwf

local must be an existing run configuration appropriate for that workflow. Test on a clean target environment so missing drivers or development-machine paths are caught before handover.

If you need remote execution, Hop Server is another option; it runs workflows and pipelines through remote run configurations.

For protecting the SQL/JavaScript, the important boundary is who administers the execution machine. File permissions can restrict ordinary users, but packaging the project in a container does not hide its contents from the host administrator. If the client controls that machine, do not treat the deployed logic as confidential from them. If keeping the implementation private is essential and the client's data requirements allow it, keep that part on infrastructure you control and expose a limited API instead.

AI-generated answer; references checked against the linked Apache Hop documentation. The example has not been executed in your environment.

[mattcasters]

Specifying the environment is fine. The project name is implied. The project folder is automatically derived.
This simplified things to:

./hop run -e production -r local -f main.hwf