Hera Pipelines Runtime

The Hera runtime allows you to run workflows within the platform. More specifically, Hera represents a workflow orchestration in Python powered by Argo Workflows in Kubernetes.

Argo workflows are defined as Kubernetes Custom Resource and represent a composition of a Kubernetes Jobs with data and parameter passing support.

Hera usess Python-based DSL to define the execuition tasks of a workflow that is then transformed into Argo workflow specification and may be executed through the platform.

In addition to Here SDK and its DSL, we provide an extension of the platform SDK to run single functions of other runtimes directly as nodes of the Hera workfflow.

Here runtime function definition is therefore composed of the Hera pipeline code in Python. The two actions supported by the runtime are

• build action necessary to compile the Hera DSL specification into Argo Workflow definition. Note that this action should be always executed before the actual pipeline execution.
• pipeline action that performs the actual instantiation and execution of the workflow within the platform using Argo Workflow operator. It is possible to pass the parameters to the pipelines.

The following figure represents the execution graph of an example worklfow in the platform:

Hera runtime replaces the Kubeflow Pipeline runtime for the same Argo workflow-based functionality.

Management with SDK

Check the SDK Hera runtime documentation for more information.