Which Kubernetes-native tool is used to define and run complex batch jobs and workflows, enabling parallel execution and dependencies?

• A. Kubeflow Pipelines
• B. Argo Workflows
• C. Apache Airflow
• D. Tekton Pipelines

Answer: (B)

Argo Workflows is a Kubernetes-native workflow engine that lets you model complex batch jobs as workflows built from a directed acyclic graph of steps. Each step runs in a container, and the Argo controller schedules them on the cluster, allowing multiple steps to run in parallel where there are no dependencies and ensuring proper sequencing where there are. Workflows are defined as Kubernetes custom resources, so they live alongside other Kubernetes resources and can be managed with standard tooling like kubectl. This combination of DAG-based execution, parallelism, and dependency handling makes it ideal for long-running data processing, ML pipelines, and any multi-stage job that benefits from explicit dependencies.

Kubeflow Pipelines targets ML workflows and sits atop Kubernetes, offering a higher-level platform for ML pipelines. Apache Airflow is a general-purpose workflow orchestrator that can run on Kubernetes but is not inherently Kubernetes-native. Tekton Pipelines focuses on CI/CD pipelines within Kubernetes. Argo Workflows uniquely emphasizes rich, Kubernetes-native DAG-based batch workflows with robust parallel execution and dependency management.