This post was written with the help of Yannick Struyf, Senior Product Manager Cloud Native at Nutanix.

Unplanned interruptions of applications results in time and financial losses for your organization, along with the challenge of restoring normal operations.

The Red Hat OpenShift 4.15 release used on the Nutanix Cloud Platform now enables you to increase the resiliency of your OpenShift cluster  and ensure the continuous operation of your applications during a disaster. Keep reading to discover how the Failure Domain feature elevates your OpenShift deployment on Nutanix infrastructure.

Distribute OpenShift nodes across Nutanix Prism Element clusters

Achieving fault-tolerant deployments is now possible with the utilization of multiple Prism Element clusters in the Nutanix environment, which are Nutanix logical entities that act as failure domains.

By strategically distributing the control plane and compute nodes across multiple Prism Element clusters, you can improve the high availability of your OpenShift environment. This multicluster (as in multiple Prism Element clusters) approach ensures that if a failure occurs in one  domain, the remaining Prism Element instances continue to provide support to the OpenShift machine pools.

All of this is possible during and after the installation of OpenShift. This architectural resilience safeguards your system against potential disruptions and fosters a more robust and reliable infrastructure for your OpenShift deployment.

Let’s see this graphically:

The above diagram shows an OpenShift cluster with failure domains for the OpenShift nodes  across three Prism Element clusters.

Adding the configuration

To distribute the control plane and compute nodes across Prism Element clusters at the time of installation, the installer is instructed  to do so in the install-config.yaml file:

platform:
nutanix:
  failureDomains:
  - name: failure-domain-1
    prismElement:
      uuid: ed4e151b-4e8d-4567-a4ee-7500ea4ed50b
    subnetUUIDs:
    - 8db8a556-7e36-446f-aa2d-ec10915b11c8
  - name: failure-domain-2
    prismElement:
      uuid: 88b721c0-a791-47bd-a03c-ab509de45b57
    subnetUUIDs:
    - 964b3373-a7a7-4dc8-a1b0-ebd03b8d8253
  - name: failure-domain-3
    prismElement:
      uuid: 17f15783-d4d9-4b7d-8efc-5d4f51b9b83c
    subnetUUIDs:
    - 850f08f7-0aa2-48af-9243-9efbad0b8a6b
  - name: failure-domain-4
    prismElement:
      uuid: ed4e151b-4e8d-4567-a4ee-7500ea4ed50b
    subnetUUIDs:
    - e19a2161-7ced-4865-9126-51d514172dc3
  - name: failure-domain-5
    prismElement:
      uuid: d3451043-fb0a-4e71-979f-6a23f9b9c2e3
    subnetUUIDs:
    - f342e8be-267a-4855-8e60-274a1978341e
controlPlane:
nutanix:
  failureDomains:
    - failure-domain-1
    - failure-domain-2
    - failure-domain-3
compute:
nutanix:
  failureDomains:
    - failure-domain-3
    - failure-domain-4
    - failure-domain-5

The  example shows  five failure domains, each mapping to a different Prism Element cluster.

The first three failure domains each host one of the three control plane nodes. Note that the control plane nodes must be on the same subnet so they each have the same CIDR in Prism Element, for example 192.168.1.0/24.

Failure domains 3-5 host the compute nodes. These failure domains can have different CIDRs, providing extra flexibility for the user to install a Nutanix cluster and place compute nodes on different subnets.

In Summary

The Failure Domain feature on the Nutanix Cloud Platform helps to increase the resiliency of your Openshift clusters. To learn more about the details and to start using it right away read the official documentation for install-time configuration  and post-installation configuration.