Skip to main content

Create a new H2O Engine

This guide walks you through creating a new H2O Engine via the H2O AI Cloud web interface.

What is H2O-3?​

H2O-3 is an open-source machine learning platform that provides:

  • Distributed machine learning algorithms
  • AutoML capabilities for automated model building
  • Support for popular algorithms (GLM, Random Forest, GBM, Deep Learning, etc.)
  • Integration with R, Python, and other programming languages
  • Scalable data processing and model training
  • Model deployment and serving capabilities
  • Interactive web-based Flow interface for data science workflows
Learn More

For more information about H2O capabilities, features, and use cases, see the official H2O documentation.

Prerequisites​

Before creating an H2O Engine, ensure you have:

  • Access to H2O AI Cloud web UI. See Access AI Engines for details.
  • Appropriate permissions allocated to your workspace. You need at least the H2O Engine User role (enginemanager-h2o-engine-user) assigned in the target workspace. See RBAC roles and permissions for the full role breakdown and Engine profile security configuration for how administrators configure profiles and versions.
  • (Optional) Resource requirements: H2O engines can be configured for various workloads. Consider your data size and computational needs when configuring resources.

Create a new H2O Engine​

To set up an H2O Engine using the H2O Cloud user interface, follow these steps:

Step 1: Navigate to Create AI Engine​

On the AI Engines page, click Create AI Engine.

create ai engine

Step 2: Select Engine Type​

On the Engine Selection tab, select H2O.

select h2o engine

Step 3: Configure Engine Details​

On the Create AI Engine > Select Engine > Engine Configuration page, provide the following information:

  1. Display Name: Enter a descriptive name for your H2O engine.

  2. Engine ID: The system automatically generates a unique ID based on your display name. However, it is recommended to use a secure ID for your use case. Click the Edit (Pencil icon) icon to customize it.

  3. Profile: Select a Profile from the dropdown. The profile determines the allowed ranges for CPU, GPU, memory, node count, and timeout settings. For more information, see H2O Engine profiles.

  4. Dataset Parameters: Configure dataset-specific parameters from the dropdown. Options in dropdown include:

    • Custom for Raw Data
    • Custom for Compressed Data
    • This can be set to "Not set" for general use cases.

  5. Version: Select the H2O version from the dropdown. The latest version is recommended unless you have specific requirements.

Engine Configuration - Display Name and ID settings

Display Name and Engine ID Guidelines

Display Name:

  • Can include special characters and alphanumeric characters
  • Character limit: 63 characters
  • Use a clear, descriptive name for easy identification

Engine ID:

  • Must start with a lowercase letter
  • Must end with a lowercase letter or digit
  • Can contain only lowercase letters, digits, or dashes (-)
  • Min length: 1 char, Max length: 63 chars
  • Cannot start or end with a dash (-)

Step 4: Configure Resources​

Configure the computational resources for your H2O engine:

  1. CPUs per Node: Set the number of CPU units per node
  2. GPUs per Node: Configure GPU units per node if needed
  3. Number of Nodes: Configure the number of H2O nodes for distributed processing
  4. Memory per Node: Allocate memory in GiB per node
Important
  • The default values and range for CPU, GPU, Memory, and Node Count are set by administrators. Contact your administrator or the H2O support team for more details.
  • H2O supports distributed processing across multiple nodes for large-scale machine learning tasks.
  • Resource allocation is calculated per node, so total resources = per-node values × number of nodes.

Step 5: Configure Timeout Settings​

Expand the Timeout Configuration section to set:

  1. Max Idle Time (Hours): Set how long the engine can be idle before automatically pausing.
  2. Max Up Time (Hours): Set the maximum duration an engine can run before automatically terminating.
Timeout Best Practices
  • Set appropriate idle time based on your usage patterns
  • Consider cost optimization vs. convenience
  • Longer idle times reduce restart overhead but increase costs
  • For H2O engines, consider the time needed for model training and data processing

Step 6: Create the Engine​

  1. Review all your configuration settings.
  2. Click Create to create the H2O engine.
  3. The engine will be provisioned and appear in your AI Engines list.

Create H2O Engine

Feedback