Getting Started With Kubeflow on Google Cloud Platform
Google announced Kubeflow in 2017 as a portable and scalable ML stack for Kubernetes. In this post, you'll learn the easiest way to get started with Kubeflow on GCP.
Join the DZone community and get the full member experience.Join For Free
The Kubeflow project was announced back in December 2017 and has since become a very popular machine learning platform with both data scientists and MLOps engineers. If you are new to the Kubeflow ecosystem and community, here’s a quick rundown.
Kubeflow is dedicated to making deployments of machine learning (ML) workflows on Kubernetes simple, portable and scalable. In a nutshell, Kubeflow is the machine learning toolkit for Kubernetes. As such, anywhere you are running Kubernetes, you should also be able to run Kubeflow. Use cases for Kubeflow include:
- Deploying and managing a complex machine learning system at scale
- Experimentation with training a machine learning model
- End to end hybrid and multi-cloud machine learning workloads
- Tuning the model hyperparameters during training
- Continuous integration and deployment (CI/CD) for machine learning
Good stuff, huh? Well, if you dig into Kubeflow a little bit, you will quickly discover that it is a collection of distinct projects like Katib, Kale, KFServing, Pipelines, and more - with each component providing essential functionality in the workflow.
So, now you may be asking yourself, “What’s the easiest way to get started with Kubeflow on GCP, with the least amount of hassle?”
Kubeflow up and Running on GCP in Just Minutes
Amazon rates MiniKF to be the fastest and easiest full-fledged Kubeflow deployment. It is hands down the simplest way to get started with Kubeflow on Google Cloud Platform (GCP).
Check out the short video below to see just how easy an installation of MiniKF on GCP is.
This video demonstrates the following steps required to set up and run Kubeflow on GCP:
- Launch a MiniKF VM on GCP
- Deploy all the necessary components including Kubernetes, Kubeflow, Kale, and Rok
- Bring up the Kubeflow UI to start your first project
With that, you are done! With only a few clicks, you are ready to run experimentations and Kubeflow Pipelines.
Published at DZone with permission of Jimmy Guerrero. See the original article here.
Opinions expressed by DZone contributors are their own.