Platform Engineering on Kubernetes

Adopting Kubernetes is hard, especially in an organization with several teams, multiple cloud providers and different technology stacks. This book shows how to build custom platforms on top of Kubernetes using open-source tools and emerging best practices from the Kubernetes community, so that development teams can stay focused on delivering software. Kubernetes handles orchestration, but efficient cloud-native delivery also needs a solid build pipeline, a package manager and distribution mechanism, and APIs that lower the cognitive load on teams.

The book opens by defining what makes up a good Kubernetes-based platform, then introduces the tools needed to shape one around your organization's requirements. Readers learn the principles of platform engineering as they apply to Kubernetes, and how to evaluate and adopt open-source projects for domain-specific platforms. They also learn to create platform APIs that help teams release more software efficiently, reduce the cognitive load of the platform, and measure platform initiatives with established software delivery metrics.

The tooling covered includes Helm, Tekton, Dagger and Argo CD for packaging, versioning, distributing and deploying, and Crossplane for a multi-cloud infrastructure strategy. Knative Serving and Argo Rollouts handle progressive upgrades, and Dapr provides standard application-level APIs for development teams. Hands-on examples and detailed code guide each step. The chapters move from the rise of platforms on Kubernetes and cloud-native application challenges to building and deploying services, multi-cloud infrastructure, building a platform, shared application concerns, enabling teams to experiment, and measuring the platform. By the end, readers can create a complete platform that is not tied to a specific cloud provider or vendor.

The book is for developers and software architects who already know the basics of containers and Kubernetes. It includes a foreword by Jared Watts. The author, Mauricio Salatino, is a Dapr open-source contributor, a member of the Knative Steering Committee and co-lead of the Knative Functions working group.