Who is your digital transformation really serving - your customers, or your shareholders? Use these exercises to determine where and how to shift focus
5 open source projects that make Kubernetes even better
Open source projects bring many additional capabilities to Kubernetes, such as performance monitoring, developer tools, serverless capabilities, and CI/CD workflows. Check out these five widely used options
Tekton spun off from the Knative project, where it was used to automate that project’s CI/CD pipeline. Because Tekton isn’t itself tied to serverless architectures, it made more sense to be organized as a separate project under the CNCF. Tekton is designed to provide an opinionated way to do continuous delivery specifically with Kubernetes. Tekton Pipelines run on Kubernetes, have Kubernetes clusters as a first-class type, and use containers as their building blocks.
Tekton is focused on cloud-native concepts like immutable container images and incorporates advanced deployment patterns like rolling, blue/green, canary deployment, or GitOps workflows. Like other newer CI/CD tools like Jenkins X, Tekton is optimized around the features and priorities of cloud-native application development, even if that means being somewhat less general-purpose than older CI/CD tools.
At a higher level in the software stack, there are also the frameworks needed to write new classes of applications, such as in the machine learning (ML) space.
Kubeflow was introduced by Google in 2017 with the stated intention of “making using machine learning (ML) stacks on Kubernetes easy, fast and extensible.” Kubeflow primarily focuses on the simplified deployment of upstream ML projects such as Jupyter, Argo Pipelines, and Istio (or curated/customized versions thereof) on Kubernetes. It was originally focused on TensorFlow (hence the name) but has since added support for other ML frameworks such as PyTorch. New ML services (such as Katib and KFServing) are now also being developed under the Kubeflow organization.
Kubeflow also serves as an upstream project for other projects using Kubernetes for ML workflows. For example, Open Data Hub inherits from upstream efforts such as Kafka/Strimzi and Kubeflow; it lets data scientists select from tools such as Jupyter notebooks, TensorFlow, scikit-learn, Apache Spark, and more for developing models.
Other Kubernetes-related projects
We’ve only scratched the surface of open source projects that can be used with Kubernetes. This short list could also have instead included a registry like Quay, the many cloud-native projects related to logging and data analysis, security and compliance scanning, developer workspaces, KubeVirt (a virtualization API for Kubernetes), or visualization tools like Grafana.
It can all be a bit intimidating, to be honest. In practice, most organizations will gravitate to container platforms that integrate many of these projects for you. But it’s still nice to know there’s so much out there for when you do need just the right tool for some important task.
[ Read also: OpenShift and Kubernetes: What's the difference? ]