Book Image

End-to-End Automation with Kubernetes and Crossplane

By : Arun Ramakani
Book Image

End-to-End Automation with Kubernetes and Crossplane

By: Arun Ramakani

Overview of this book

In the last few years, countless organizations have taken advantage of the disruptive application deployment operating model provided by Kubernetes. With Crossplane, the same benefits are coming to the world of infrastructure provisioning and management. The limitations of Infrastructure as Code with respect to drift management, role-based access control, team collaboration, and weak contract make people move towards a control-plane-based infrastructure automation, but setting it up requires a lot of know-how and effort. This book will cover a detailed journey to building a control-plane-based infrastructure automation platform with Kubernetes and Crossplane. The cloud-native landscape has an overwhelming list of configuration management tools that can make it difficult to analyze and choose. This book will guide cloud-native practitioners to select the right tools for Kubernetes configuration management that best suit the use case. You'll learn about configuration management with hands-on modules built on popular configuration management tools such as Helm, Kustomize, Argo, and KubeVela. The hands-on examples will be patterns that one can directly use in their work. By the end of this book, you'll be well-versed with building a modern infrastructure automation platform to unify application and infrastructure automation.
Table of Contents (16 chapters)
1
Part 1: The Kubernetes Disruption
4
Part 2: Building a Modern Infrastructure Platform
10
Part 3:Configuration Management Tools and Recipes

Defining API boundaries

We expect platform engineers to compose all the infrastructure and application automation concerns in XR APIs. How to define an API boundary is a bit tricky. It’s because many conflicting trade-off points are influencing the API boundaries. Let’s start with the fact that we wanted to compose every resource required for an application and its infrastructure in a single composition. Here are some considerations that will not allow us to do that:

  • There would be many security and architecture policies that need to centralize. We cannot add them again and again in multiple compositions.
  • Some resources may have compliance requirements and must be composed separately for audit purposes.
  • Overly big compositions are difficult to read, understand, debug, and refactor.
  • Testing a bulk composition is difficult.
  • Every application will require some customization to the infrastructure recipe, provided we have a bulk composition. We will...