Multi-Container Environments#

You can use multi-container environments in AI Workbench through the Compose Applications feature.

It is based on the compose specification and features that are described in the Docker Compose documentation.

Overview of Compose in Workbench#

You can add a multi-container application to a Workbench project with a compose file.

The feature works with all supported container runtimes: Docker, Docker Desktop, and Podman.

Using a multi-container application in Workbench should be straightforward.

Using a multi-container application in Workbench is easy if you use the following steps:

  • Read the relevant section of the readme to understand any process required.

  • Make sure any necessary environment variables are configured.

  • Go to Environment > Compose and start the application.

However, creating and configuring a multi-container application requires technical skill.

Creating and configuring a multi-container application requires the following skills:

  • You need to know how to read, interpret and edit a compose file.

  • (if required) You need to know how to specify the correct environment variables and secrets.

  • (if required) You need to know the proper GPUs required for the different services.

  • (if required) You should setup any required profiles to account for different GPU configurations.

Note

There may be more concerns to handle depending on the application you are using or creating.

Note

The best way to understand how multi-container environments work is to work through the exercises in the Onboarding Project.

FAQs#

Why would I need a multi-container environment?#

Multi-container environments are useful anytime you have different services or applications that are difficult or impossible to run in a single environment or container.

They’re also useful to maintain isolation and test interactions between different services. Some examples include:

  • If you are building a web app with a database.

  • If you want to use a full stack development environment.

  • If you want to do CI/CD pipeline testing locally.