Skip to main contentIBM Cloud Pak Playbook

Introduction

Openshift 4 is the base platform for all Cloud Paks described on this site. Installing it in a consistent way in both on-premises and the public cloud provides a stable landing ground for Cloud Pak deployment. This section discusses the installation of Openshift 4 on various cloud infrastructures and introduces the concepts around the Terraform automation tool. On this site, every reference to OpenShift, unless otherwise specified, will assume OpenShift 4.

As of this page update, the latest generally available version was OpenShift 4.3. OpenShift 4 is a major advance over OpenShift 3, and contains a number of significant architectural changes. The official Red Hat OpenShift documentation includes a product architecture overview that describes the structure and functions of OpenShift 4.

OpenShift provides a consistent application environment on a variety of infrastructure platforms. These platforms include public cloud providers, such as Amazon Web Services (AWS), Microsoft Azure, Google Cloud Platform (GCP), and IBM’s managed Red Hat OpenShift service (ROKS). They also include platforms traditionally thought of as on premise, such as VMware, OpenStack, and bare metal. Wherever it is installed, it provides a consistent container application runtime, APIs, and interfaces.

Since the focus of this site is on installing OpenShift 4 and the IBM Cloud Paks, installation is where we will focus.

OpenShift Installation Overview

One of the major advances in OpenShift 4 is the installation process. There are two types of installation available:

  • Installer-provisioned infrastructure (IPI), where the installation program also provisions all the infrstructure (machines, networks, storage, etc) required for the cluster.
  • User-provisioned infrastructure (UPI), where the installation program installs the cluster on infrastructure that has previously been set up by the user.

The default installation type is installer-provisioned infrastructure, which is recommended wherever possible. Using this method, the installation program works as a wizard, prompting you for configuration values that it cannot determine by itself. You are able to install either a standard cluster with preset starndard sizes and configurations, or you have the option of specifying more customized parameters. With IPI, OpenShift itself manages all aspects of the cluster, including the operating systems of the nodes. Again, the IPI method is recommended wherever possible.

User-provisioned infrastructure installation can be used in special circumstances requiring more customized infrastructure requirements. One of these requirements is a network restricted or “air-gapped” installation, where the cluster to be installed is not allowed to have internet access. In this case, the installation images can be staged on an internal network, and UPI installation can be used. If UPI installation is used because of client requirements, tools such as terraform can be used to to automate the setup of the user-provisioned infrastructure, and the installtion of OpenShift itself. This web site provides links to example terraform configurations for most of the infrastructure platforms described.