Kubernetes clusters made simple: Fast, easy Red Hat OpenShift on HPE bare metal servers

ServicesExperts · ‎04-19-2022

The deployment and management of Kubernetes clusters on bare metal has become a challenge for many organizations. HPE and Red Hat combine their technologies to provide an enterprise-grade approach to automate the build, configuration and run of Red Hat OpenShift clusters on HPE hardware.

By Nicolas Orand, Senior Chief Solution Architect, Cloud Native Computing Practice Area, HPE Advisory & Professional Services

Why bare metal for Red Hat OpenShift?

Today, most Red Hat OpenShift installations are done on virtual machines, or directly on public cloud due to its simplicity and flexibility and/or the pay-as-you-go model. With the advent of machine learning frameworks on Kubernetes, companies are looking more and more at using bare metal workers in their clusters. In the past, these deployments used to be complex and required a lot of planning around users, providers and infrastructure installers. Today – thanks to HPE and Red Hat’s set of technologies – it has become exponentially easier to automate OpenShift installation on top of bare metal servers.

This blog aims to present the diverse ways to install Red Hat OpenShift on HPE Servers (HPE Synergy and HPE ProLiant) leveraging HPE and Red Hat solutions.

One question commonly asked by customers is: Why install Red Hat OpenShift on HPE bare metal servers directly?

Bare metal provides the same flexibility as a virtual environment. Today the full infrastructure-as-code reaches the capabilities of the VM (virtual machine) provisioning services.
It can be consumed in a pay-as-you-go model thanks to the HPE GreenLake edge-to-cloud platform.
By removing the hypervisor layer, you reduce the license costs associated with the hypervisor, and you improve security by reducing the attack surface.
You can increase performance by giving direct access to hardware like GPUs (widely used in machine learning frameworks).
Customers looking to manage VMs as Kubernetes objects can do so via OpenShift Virtualization (KubeVirt).
Last but not least, Red Hat offers a competitive subscription model for bare metal nodes.

Customers can count as well on HPE Advisory and Professional Services offerings to ease the deployment and maintenance of Red Hat OpenShift on HPE Servers.

How to deploy Red Hat OpenShift on bare metal

Automating the deployment of OpenShift on bare metal servers is now a much easier endeavour. As seen in the global architecture below, the automation should go through three main stages:

Configuration/Hardware Provisioning
Cluster Creation
Cluster Customization

The first stage is the provisioning of the hardware. In this phase we leverage HPE OneView to define server profile templates, allocate free hardware and create the server profiles. It is possible to use Ansible collection or OneView SDK (Software Development Kit) to consume the API (Application Programming Interface) provided by HPE OneView. This way we can provide a list of free servers ready to install OpenShift.

The second stage is the creation of the OpenShift cluster itself. We have the choice to leverage on the OpenShift installer or the Red Hat Advanced Cluster Management (ACM) for this task.

The third stage is the cluster customization. Customization can be achieved by using a combination of Argo CD, ACM and Ansible. It will be used for all post-configuration tasks such as defining quotas, installing operators like HPE CSI (Container Storage Interface) Drivers, and configuring the container registry.

This approach allows customers to deploy Red Hat OpenShift in one hour on HPE bare metal servers (HPE Synergy or ProLiant DL) and automate the post-configuration.

Two common use cases for the automated deployment are:

1. You’re looking to do a greenfield deployment and have Red Hat OpenShift Container Platform subscriptions available.

The installation requires only 3 steps:

Use Ansible playbooks to trigger HPE OneView and provision the hardware.
Use the OpenShift installer to install the cluster in IPI (Installer Provisioned Infrastructure) or UPI (User Provision Infrastructure) mode.
Create post-configuration, including Ansible and Argo. This includes the scale-up of node, for example.

2. You’re looking to deploy multiple Kubernetes clusters in your organisation and you have Red Hat OpenShift Platform Plus subscription. Again, the installation requires only three steps:

Use HPE OneView SDK combined with Tekton pipeline to prepare the HPE hardware and update the ACM inventory.
ACM creates the cluster automatically with the IPI method.
Create the post-configuration via Ansible and Argo. This includes the CSI or OpenShift Data Foundation configuration as an example of post-configurations available.

These two use cases can be demonstrated at HPE Innovation Centers.

Each customer will have their own constraints for containers or automation. HPE Advisory and Professional Services can help you with defining your strategy, as well as the design, delivery, implementation and operation of the solution. This will definitely accelerate the adoption of these technologies in your company.

Learn more about technology services consulting from HPE Pointnext Services.

Nicolas Orand is Senior Chief Solution Architect in HPE Advisory & Professional Services’ Cloud Native Computing Practice Area. He has been with HP/HPE for 11 years, working on hybrid cloud for five years. Nicolas has three focus areas: He is a Global Cloud Native Computing Container solution development architect. He’s an EMEA Red Hat Cloud Architect for HPE Advisory and Professional Services, enjoying a great partnership with Red Hat. And he’s responsible for France Delivery for HPE hybrid cloud products. Nicolas is based in Grenoble.