Drop by the Arrikto booth S/E 53 at KubeCon Barcelona 2019. Meet the experts, and watch the new, on-prem Kubeflow Pipelines demo. Get the
t-shirt!
Today, at KubeCon + CloudNativeCon Europe 2019, Arrikto announced the release of the new MiniKF, which features the all-new Kubeflow v0.5. MiniKF is the fastest and easiest way to get started with Kubeflow. With more than a thousand downloads in the last 60 days, MiniKF is becoming the de-facto Kubeflow distribution for laptop and desktop use.
MiniKF installs with just two commands. Then, you can start experimenting, and even run complete Kubeflow Pipelines use cases. Portability is key, so you can move your training or model deployment to a Kubeflow cloud deployment with one click, without having to rewrite anything. MiniKF runs on all major operating systems (Linux, macOS, Windows).
What’s new in MiniKF
- Kubeflow v0.5
- New Notebook Manager UI (contributed by Arrikto)
- Support for local and on-prem Kubeflow Pipelines
- Seamless Notebook integration with Pipelines (with Arrikto’s Rok)
The new MiniKF delivers a simplified model building experience for data scientists and dramatically improves the workflow to create and run a Kubeflow Pipeline. MiniKF comes with all the features of Kubeflow v0.5, plus the ability to run end-to-end Kubeflow Pipelines locally starting from your Notebook. With MiniKF and Rok your workflows just became truly hybrid. Take an immutable snapshot of your local Notebook, and its data (using Arrikto’s Rok data management platform), and continue working, or reproduce on a public cloud provider like GCP. Use Kubeflow Pipelines to automate the procedure and run steps efficiently accessing local data. Everything happens from inside your Notebook! MiniKF comes with a free Rok license, so that you can enjoy all of Rok’s advanced data management capabilities.
Here is an example of how to run an end-to-end Kubeflow Pipeline locally, on MiniKF, starting from a Jupyter Notebook.
Why we started MiniKF
Arrikto is a leading code contributor to Kubeflow, and is driving the Community’s efforts to deliver efficient data management solutions for multi-user and on-premise workflows. Kubeflow is dedicated to making deployments of machine learning (ML) workflows and overall MLOps on Kubernetes simple, portable and scalable. The fact is that data scientists use their laptop a lot! The vast majority of data science exploration and experimentation starts locally from the laptop. So, the more we interacted with the Kubeflow community, the more it became clear that we needed a way to make Kubeflow easily accessible, and packaged in a simple way to run locally.
If you are a Kubernetes user, you can think of MiniKF being to Kubeflow, what Minikube is to Kubernetes.
We believe MiniKF will help democratize access to ML, lowering the barrier to entry, and allowing more people to spin up a pre-configured dev environment, so they can start experimenting locally, fast, and then scale-up to a full, distributed operation on the cloud. With scaling-up becoming the next step of a single workflow, rather than a completely distinct hassle.
Availability
MiniKF is already available and completely free of charge. Visit the MiniKF page to download it and start using it:
See also the initial, official announcement on the Kubeflow blog:
We welcome your feedback, suggestions, bug reports, and feature requests. Give it a try and tell us what you think!
Join the #minikf channel on Kubeflow Slack