Transforming WebSphere to Open Liberty on OpenShift: An Operations point of view — Part 1

Eric Herness
AI+ Enterprise Engineering
4 min readMay 10, 2021

This article written by Eric Herness and Karri Carlson-Neumann

Introduction

Enterprises are pushing forward to newer technologies, such as containerized workloads running on kubernetes platforms such as OpenShift Container Platform (OCP). Their ability to sustain existing workloads on traditional middleware, such as WebSphere Application Server Network Deployment (WAS ND) environments, faces logistical challenges. Their very skilled runtime staff of operators and administrators may have already been downsized to “just keep the lights on” staffing numbers. Replacing people with this deep knowledge is becoming difficult.

No enterprise has a truly cookie-cutter implementation of WAS ND, as various services, resources, and parameters have been tuned or customized as per requirements. Some enterprises are better than others about having documented everyday procedures and special one-offs. There are usually a few people with some very deep, very custom knowledge about the WAS ND environments and workloads, and they are extremely valuable.

When it comes time to progress these workloads to OCP, the success of this undertaking depends a lot on the ability of the WAS admins, operators, and architects to understand how these environments, workloads, and their details change, map, and behave in OCP. This can be quite a challenge, as the initial learning curve from a WAS ND world to a containerized OCP world can be quite steep.

The goal of this series of blog entries is to help prepare administrators, operations teams, and architects for a WAS ND to Open Liberty on OCP transition. While it is impossible to contain every consideration in a reasonably-sized document, the intention is to provide a running start at a number of them.

If your organization is containerizing WAS ND applications, or considering this modernization approach, read on. If your organization is only considering full refactoring of existing applications into a microservices architecture, then the material here is far less useful, as you’ll find yourself starting from a clean slate. Our experience, at a broader level, with modernization is that a blend of techniques is the best, and containerization and repackaging need to be amongst those pursued.

The backgrounds of the authors and contributors to this series of articles includes cumulative decades of WAS ND (and WebSphere Process Server/Business Process Manager) development and architecture work. Each of us transitioned into containerized workloads and Kubernetes-based platforms a few years ago. We very specifically know the challenges that existing WAS ND admins, operations teams, and architects are facing, and can still speak in WAS ND terminology.

What will be covered

Even at a high level, this is an extensive amount of data to process. We have split the content into smaller units to be more consumable and will build on each other.

  • Part 1: Introduction — this is what you are reading now.
  • Part 2: Overview of WAS ND
  • Part 3: Overview of OCP and Open Liberty
  • Part 4: Mapping topologies
  • Part 5: Operational comparisons

In Part 2, we will overview the major terminology and concepts that you are already familiar with in your WAS ND deployments. It is assumed that you know these details quite well already. The purpose of the overview is to highlight the terminology and concepts that will be used again later in the series as we are drawing comparisons.

In Part 3, we will overview the major terminology and concepts used in OCP and Open Liberty environments. We will then compare many of these terms and concepts back to the similar idea from the WAS ND overview. The good news is that there is a pretty good mapping. WAS ND is a control plane, OpenShift provides a control plane, not equal, but solve lots of the same problems. You see where this is going. We’ll help you get a handle on this and show you where the comparisons work and show subtle differences to recognize as well.

In Part 4, we will take a high-level view of mapping WAS ND topologies to OCP topologies. In the end, we hope you will see that OpenShift (and Kubernetes in general) does account for and enable some of the crafty and sneaky things we’ve seen done in WAS configurations. I think we would say that OpenShift often does this somewhat differently, but quite elegantly in most cases.

In Part 5, we will focus on comparing a sampling of tasks performed in a WAS ND environment to the roughly equivalent tasks for Open Liberty applications in an OCP environment. We will hold off on providing too much opinion here in this section, but again we think you will see some of the reasons there is great value in containerizing WAS applications into OpenLiberty and running them on OpenShift.

Final Words

We know that the unmodernized 80% of workloads that everyone likes to speak of contain some of the most mission critical business functions. We know that many of these are running on WebSphere. There is a lot to think about when considering and planning for modernization of these crown jewels. While the developers have a role, the rest of the team that makes these applications hum and deliver on their SLAs are also key cogs in ensuring a successful modernization journey. Go here to find Part 2.

--

--