Detailed Description
Many Kubeflow users have raised concerns about its complexity, particularly the number of control plane dependencies required for the default installation #2451.
Given recent discussions around Helm Charts, we should explore ways to reduce the number of default dependencies to make it easier for users to try Kubeflow projects. While these dependencies are essential for a fully featured, end-to-end ML platform, they are not strictly required for a Kubeflow MVP. We can always provide documentation guides/instructions for distributions and users on how to enable additional dependencies.
I want to start this umbrella issue to see how we can simplify the default Kubeflow installation in kubeflow/manifests, allowing users to install it more easily via Helm Charts or Kustomize Manifests and quickly realize the AI/ML value of the Kubeflow ecosystem.
cc @kubeflow/kubeflow-steering-committee @kubeflow/wg-notebooks-leads @kubeflow/wg-data-leads @kubeflow/wg-automl-leads @kubeflow/wg-pipeline-leads @kubeflow/wg-training-leads @kubeflow/wg-manifests-leads @brsolomon-deloitte @ruckc @doncorsean @kubeflow/release-team
@chasecadet @franciscojavierarceo @indemnifyai @kromanow94 @akgraner
Detailed Description
Many Kubeflow users have raised concerns about its complexity, particularly the number of control plane dependencies required for the default installation #2451.
Given recent discussions around Helm Charts, we should explore ways to reduce the number of default dependencies to make it easier for users to try Kubeflow projects. While these dependencies are essential for a fully featured, end-to-end ML platform, they are not strictly required for a Kubeflow MVP. We can always provide documentation guides/instructions for distributions and users on how to enable additional dependencies.
I want to start this umbrella issue to see how we can simplify the default Kubeflow installation in
kubeflow/manifests, allowing users to install it more easily via Helm Charts or Kustomize Manifests and quickly realize the AI/ML value of the Kubeflow ecosystem.cc @kubeflow/kubeflow-steering-committee @kubeflow/wg-notebooks-leads @kubeflow/wg-data-leads @kubeflow/wg-automl-leads @kubeflow/wg-pipeline-leads @kubeflow/wg-training-leads @kubeflow/wg-manifests-leads @brsolomon-deloitte @ruckc @doncorsean @kubeflow/release-team
@chasecadet @franciscojavierarceo @indemnifyai @kromanow94 @akgraner