I help teams migrate their applications onto Red Hat OpenShift, so I can't help but notice patterns and considerations that arise regarding the migration process. Such operations have many domain-specific factors, but in regards to getting the applications up and running on OpenShift, there appear to be several common patterns that teams use to migrate successfully.
I've found the following are useful effort breakpoints:
- Proof of concept.
- Set up continuous integration and continuous delivery (CI/CD) for a single environment.
- Set up CD for multiple environments.
In this series, I will break down the work involved in each of these stages and use the
bookinfo application from the Istio project to demonstrate. I selected
bookinfo because it is an existing sample microservices application that I can deploy from scratch onto OpenShift. Additionally,
bookinfo is polyglot (the microservices are each written in different languages), which allows me to showcase my thought process by migrating different types of services across a spectrum of programming languages.
I forked the
bookinfo subdirectory from Istio using the process described by GitHub: Splitting a subfolder out into a new repository. You can find my fork at
rh-tstockwell/bookinfo. The master branch is a direct fork from
istio/istio, while the development branch contains the changes made in this series.
For this series, I assume you have at least beginner OpenShift, Kubernetes, and container knowledge. Additionally, you will need developer access to an OpenShift cluster. If you do not have one available to you, you can try it out for free (see Get started with OpenShift).
You will also need to install the
oc client and successfully log into your cluster with standard developer privileges. If you followed one of the guides from the get started page, your guide should have instructions for you.
I primarily use the
oc client in this series (I generally prefer the command line), but you can make all of these changes using the GUI as well.
Proof of concept
First, the goal is to get your application up and running quickly to act as a proof of concept. The aim is to understand the complexity involved in your undertaking as early as possible to de-risk what you can upfront.
My rules of thumb for this stage are, when possible:
- Spend as little time as possible getting your apps to work.
- Use standard s2i images or templates (learn why here).
- Avoid code changes.
Note: From experience, it is Hard Work™ supporting drifting codebases during a platform migration, so we try to avoid this issue where we can.
In Part 2 of this series, I will demonstrate how easy it is to get
bookinfo up and running on an OpenShift cluster using
Set up CI/CD for a single environment
The second goal is to be able to make changes to your application through a simple CI/CD pipeline. This pipeline allows you to evolve your application faster, and with more dependability and assurance.
- Runs your defined tests.
- Deploys new application code on demand.
- Deploys updates to Kubernetes resources required by the application.
Ideally, this process should work from and to any possible state.
Once you have a completed pipeline, I recommend cleaning up any technical debt that resulted from getting your apps working. You should also implement anything that you left out of the proof of concept that you require for a working, production-ready application (from an application perspective, not necessarily from an ops perspective). I purposefully delay dealing with these issues until we have a working pipeline so we can use the benefits of the pipeline when implementing these more difficult features.
Set up continuous delivery for multiple environments
Lastly, we need to reconfigure the CD pipelines we created in the previous step to add the ability to deploy to multiple environments. I like to keep this step separate from the previous one as it usually involves extra steps, configuration, and templating of the k8s resources. You can also implement this step in many ways.
In Part 4 of this series, I will show how to implement a basic multi-environment CD pipeline in Jenkins with no external software.
Now, I haven't tried to cover every single aspect of running an application on OpenShift, but here a few things that I haven't covered that you might want to investigate once you have reached this stage in your journey:
- Manage your k8s resources with GitOps (you should already be part-way there, as the pipelines already source the k8s resources from version control).
- Improve your secrets management with something like HashiCorp Vault.
- Set up a persistent storage/backup/redundancy lifecycle (which is dependant on your specific environment and needs).