What is Argo?
Argo Workflows is an open-source and container-native workflow engine that helps orchestrate parallel jobs on Kubernetes. It is implemented as a Customer Resource Definition of Kubernetes.
Argo enables defining workflows with every step in the pipeline as a container. It allows creating multi-step workflows with a sequence of tasks and mapping the dependencies using DAGs.
Argo supports building portable and version-controlled pipelines for compute-intensive data science jobs. It contributes to the Kubernetes ecosystem with better accessibility, CI/CD support, and advanced Kubernetes deployment strategies.
How Does Argo Help?
Argo workflows provide a mechanism for feeding and tending a Kubernetes cluster. It helps specify, schedule and coordinate the execution of intricate task pipelines and applications on Kubernetes. Argo provides the following benefits.
- It is specially designed for containers eliminating the limitations of server-based environments and legacy virtual machines
- Supports orchestration of complex parallel jobs on Kubernetes with adequate dependency tracking
- Helps orchestrate complex, distributed applications
- Facilitates time and event-based execution of workflows
- Go modules support
Key Features of Argo
Cloud agnostic
Argo is a lightweight, cloud-agnostic workflow engine that works with any Kubernetes cluster. It is designed to be configurable and operates across environments.
Fully featured UI
Argo offers a fully-featured, robust, and reliable UI to support Argo Events, DAGs, workflow log viewer, embeddable widgets, configurable “Get Help” button, and configurable link buttons.
Controller-High availability
Argo v3.0 brings a hot-standby workflow controller feature for quick recovery and high availability. It leverages the Kubernetes leader election feature for faster crash recovery.
Key-Only artifacts
Argo provides a default artifact repository and key-only artifacts that work together. Default artifact repository is configured for user namespace, avoiding explicit repository definition for each workflow. Argo workflow specification requires “key-only” fields, and non-key fields are inherited.
New API endpoints
Argo Workflows 3.0 brings brand-new API endpoints for Argo events. The event-flow page helps understand how event sources and sensors are linked, as well as trigger-based linking in workflows.
New repository location
Argo workflow repository is renamed as argo-workflows. The new name indicates that it is the repo for Argo Workflows and not the entire Argo Project.
