Skip to content
Tutorials Center
My learning

Topics

Lesson 2 of 37 ยท Getting Started

Editions and Ways to Run OpenShift

Compare OpenShift Container Platform, OKD, managed services and free options such as the Developer Sandbox and OpenShift Local, and pick the right way to learn.

"OpenShift" is a family of products rather than one download. Before you start, you should know which flavors exist, who runs the control plane in each, and which one is the cheapest way to practice. Most of what you learn in this track works on all of them.

The main flavors

OpenShift Container Platform (OCP) is the commercial, supported product. You install it yourself on your own hardware or on a cloud account, and you run and upgrade it (with the platform's help). It is what most enterprises mean by "OpenShift".

OKD is the community upstream of OpenShift. It is free, closely tracks OCP features, and is what you would use to explore the code or run a lab without a subscription. It is not covered by Red Hat support.

Managed OpenShift means a provider operates the cluster for you. Well-known examples are Red Hat OpenShift Service on AWS (ROSA), Azure Red Hat OpenShift (ARO) and Red Hat OpenShift on IBM Cloud. Your team uses the cluster; the provider and Red Hat's site reliability engineers keep the control plane healthy.

OpenShift Dedicated is a managed offering operated by Red Hat. Offerings and names change over time, so always check the provider's current documentation before you plan a purchase.

flowchart LR
  U["OKD<br/>community, free"] --> P["OCP<br/>supported, self-managed"]
  P --> M["Managed<br/>ROSA, ARO, IBM Cloud"]

Figure: the same platform, with more operations handled for you as you move right.

Free ways to learn

You do not need to buy anything to practice.

  1. Developer Sandbox for Red Hat OpenShift. A free, shared, time-limited cluster you sign into with a Red Hat account. You get a couple of ready projects, the web console and the oc CLI. It is the quickest start for the developer chapters of this course. Sandbox clusters have small quotas and you cannot act as cluster administrator.
  2. OpenShift Local (the successor of CodeReady Containers, command crc). It runs a single-node OpenShift cluster in a virtual machine on your laptop. Expect to give it several CPU cores and around 10 GB or more of memory. You get cluster-admin rights, so it is the best option for the administration chapters.
  3. Single-Node OpenShift (SNO). A full install on one server. Often used for edge and labs.
  4. Hosted learning labs. Red Hat and partners run time-boxed interactive labs in a browser.

Use the Developer Sandbox for chapters about projects, builds, routes and pipelines. Switch to OpenShift Local or a lab cluster for chapters about nodes, upgrades, Operators and cluster-wide security, because those need cluster-admin.

Installation methods in one page

On your own infrastructure there are two broad install styles.

  • Installer-provisioned infrastructure (IPI). You give the installer cloud credentials; it creates the machines, networks and load balancers, and then the cluster. Least work, most automation.
  • User-provisioned infrastructure (UPI). You build the machines and networking; the installer then configures OpenShift on them. More control, more work.
  • Agent-based and Assisted installers. Boot a discovery image on bare metal and the installer guides the rest. Common for data centers.

The installer program is called openshift-install. A minimal flow looks like this (read-only: it needs credentials and real infrastructure):

Terminal
openshift-install create install-config --dir mycluster
openshift-install create cluster --dir mycluster --log-level=info

The first command asks questions (platform, region, base domain, pull secret) and writes install-config.yaml. The second builds the cluster, which typically takes a while, and leaves credentials in mycluster/auth/, including the kubeconfig file and the initial kubeadmin password.

What sizes look like

A production cluster usually has three control-plane nodes (for etcd quorum) and at least two or three worker nodes. Learning clusters are far smaller. The stand-in below checks whether a proposed layout is sensible. The rules are simplified teaching rules, not Red Hat's sizing guide.

Python Stand-in: is this cluster layout sensible?

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Picking an option

The right answer depends on your goal. This stand-in turns the advice into a tiny decision function you can adapt.

Python Stand-in: choose a way to learn

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Versions and lifecycle

OpenShift 4 releases a new minor version (4.x) several times a year. Each minor version is supported for a limited period, and some versions are designated for longer support. Upgrades move one minor version at a time along tested paths, and the cluster tells you which targets are available. You will meet this properly in the administration chapter.

Release names and support periods are published by Red Hat and change, so this course avoids quoting specific dates.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Which option is the community upstream of OpenShift Container Platform?

    OKD is the free community distribution that OCP is derived from. ROSA is a managed service, crc is OpenShift Local and SNO is a one-node install.
  2. 2. Which free option gives you cluster-admin rights on a cluster running on your own laptop?

    OpenShift Local runs a single-node cluster on your machine and you are the administrator. The Sandbox is shared and does not give cluster-admin.
  3. 3. In an IPI install, who creates the machines and load balancers?

    Installer-provisioned infrastructure means the installer uses your cloud credentials to create the infrastructure.
  4. 4. Why do production control planes normally have three nodes?

    etcd uses a majority vote; with three members it keeps working if one fails.

Summary

  • OCP is the supported product, OKD the community version, and managed services such as ROSA and ARO run the control plane for you.
  • The Developer Sandbox is the quickest free start; OpenShift Local gives you cluster-admin on your laptop.
  • IPI automates infrastructure creation; UPI means you prepare it.
  • Production clusters use three control-plane nodes plus workers.
  • Always check current Red Hat documentation for versions and support periods.

Next: The oc CLI and the Web Console