Locations

This is a high-level conceptual overview of NVIDIA AI Workbench Locations. For a guide to features, see the corresponding How-To topic(s) (Desktop and CLI) and for detailed reference information, see the Deep Dive topics.

  • A Location is a single system with AI Workbench installed. It can be local or remote.

  • You must install AI Workbench on your primary local system, e.g. laptop.

  • You can add multiple remote locations, and each gets a unique name and has its metadata like IP address.

  • Projects are created or cloned into locations, and you manage and work in them from the Workbench UI running locally.

  • Projects are portable and reproducible across locations with minimal need for adjustment.

AI Workbench Application Components

Workbench Service

A single binary that provides the core application. It is installed on your primary local system, as well as on remote systems.

Credential Manager

A single binary that handles authentication for the various integrations. It is installed on your primary local system. On remote systems, it is installed but not used.

Command-line Interface (CLI)

A single binary that provides a CLI. It is installed on your primary local system, as well as on remote systems.

Desktop App

An Electron app that is installed on your primary local system, e.g. your laptop, for a graphical user interface. It is not installed on remote systems.

Local vs Remote Locations

The main distinction is between your local and remote Locations.

  • Local (Local AI Workbench): This is your primary location, typically a laptop, and it is named local in the UI. It has all components of the AI Workbench application installed, including the Desktop App. You have only one, and it is effectively permanent until you get a new laptop.

  • Remote (Remote AI Workbench): This is a system that you can access via an SSH connection. It has all application components installed except for the Desktop App. To use a Remote Location, you first need to add it to your local AI Workbench.

Adding Remote Locations to Your Local AI Workbench Client

The UI of your local AI Workbench allows you to add remote Locations, and this gives you access to the Projects in that Location. You can work in those locations to develop and compute interactively with everything running remotely. You can add more than one location, and you can delete locations as well.

Active vs Deactivated Location

A Location must be active before you can develop and compute in it. This means that the AI Workbench UI has successfully connected to the Workbench Service running in the Location.

Activating a Location

  • (Remote Locations Only) Establishes two SSH tunnels between your local system and the remote system. One for the Workbench Service and one for a proxy that all Project applications run behind.

  • Makes sure the container runtime (Podman or Docker) is running properly in the Location.

  • Makes sure the Workbench Service is running properly in the Location.

  • Pushes credentials into memory in the Location for AI Workbench to use when needed.

  • [CLI] Initiates a session connected to the Location and updates your terminal prompt to show the Location is active in the terminal.

  • [Desktop App] Opens a window for the Location.

Deactivating a Location

  • (Remote Locations Only) Closes the SSH connection.

  • [CLI] Removes Location information from the terminal prompt. If the --shutdown flag is used, it also kills the session.

  • [Desktop App] Closes the Location window and closes the Workbench Service and runtime if no other locations are open. On Mac, it closes the Workbench Service and stops Podman.

Current Limitations

  • Remote Locations support only one user. They are not multi-tenant and they don’t support concurrent users.

  • Remote Locations are currently only possible on Ubuntu 22.04. Other remote locations will be available in GA.

overview-concepts-locations-1.png

Previous Introduction
Next Integrations
© Copyright 2023-2024, NVIDIA. Last updated on Jan 21, 2024.