Kubeflow

Provides a brief overview of Kubeflow in HPE Ezmeral Unified Analytics Software.

Kubeflow is the platform to develop and deploy machine learning (ML) workflows using Kubeflow components. You can create Kubeflow pipelines, manage Katib experiments, and serve ML models using Kubeflow in a fully managed and secured unified environment provided by HPE Ezmeral Unified Analytics Software.

The external link for KServe InferenceService follows the following pattern: 
service-name.namespace.domain.com
A dot is used between the service name and its namespace, creating two subdomains.

Features and Functionality

Kubeflow in HPE Ezmeral Unified Analytics Software supports the following features and functionality:

  • Provides a seamless SSO login experience for authorization and authentication.
  • A default notebook is created with tensorflow image by using Kubeflow Notebooks. See Creating and Managing Notebook Servers.
  • A default user volume is created in Kubeflow notebooks where only the current user has access to the data stored in the user folder.
  • Kubeflow notebooks contains the shared directory with all the notebook examples that can be accessed by all the authorized users.

Kubeflow Components

The following are Kubeflow components:

To learn more, see Kubeflow documentation.