Upgrading OpenStack: why containerization and Kubernetes are the answer

Organizations running an open source private cloud know OpenStack as the engine behind their infrastructure. The software delivers VMs, network capacity, and storage, the same way a public cloud provider does. Yet the biggest challenge does not sit with the first installation. That challenge sits in maintaining, patching, and securing the system year after year. Fairbanks explains how containerization and Kubernetes support that work, and why Talos Linux fits as a natural addition.

At the start of an OpenStack environment, attention goes to configuration and connecting the system to the data center. After a few years, that attention shifts. The focus moves to upgrading, patching, and keeping the system in a secure, current state. That shift is where most of the workload for admin teams builds up.

OpenStack was designed fifteen years ago, and that design still holds up well. The architecture resembles what we now call a cloud-native application. A few stateful components exist, such as a relational database and message queues. The rest consists of Python code that runs as stateless services and offers a REST API to users.

That structure makes OpenStack fit for modern management methods, even though the software predates the term cloud-native. In the early years, OpenStack components shipped as standard packages through Linux distributions. Today, many providers offer containers for each OpenStack component, with regular updates and multiple options per component.

Containerization changes how teams manage OpenStack. Instead of separate configuration files per server, components run in containers that admins update and replace in a uniform way. That saves time and prevents drift between servers that builds up over years.

Most OpenStack environments deployed today already use containers. Some still rely on lightweight Linux containers, while others move to application containers managed in a different way. This shift brings consistency and makes maintenance predictable, regardless of the number of nodes in the data center.

Once a system runs in containers, orchestration comes into play. For that orchestration, the open source community widely picks Kubernetes. That choice is not a coincidence. OpenStack was designed for orchestration from the start, and that function now belongs to Kubernetes. Several projects work actively on running OpenStack on top of Kubernetes.

Before Kubernetes existed, separate tools emerged that each tried, in their own way, to build a stable, manageable, upgradable OpenStack environment. As a result, knowledge and expertise stayed scattered across those tools. Kubernetes brings that orchestration together in one consistent format. Running OpenStack on Kubernetes adds no extra complexity, as long as teams follow the right approach.

The challenge lies in deploying a Kubernetes cluster tailored to a single application, OpenStack itself, without translating all of OpenStack’s overhead directly into Kubernetes concepts.

Before an organization can orchestrate OpenStack through Kubernetes, it needs a Kubernetes cluster. Fairbanks uses Talos Linux for that purpose. Talos Linux is a Linux distribution built solely to run Kubernetes, also referred to as a Kubernetes Operating System.

Traditional Linux distributions offer no ready-made Kubernetes package. Admin teams spent years adding separate tools to install Kubernetes on top of an existing distribution. Talos Linux solves that by putting simplicity at the center of operating system management. It is a compact, immutable system, built for one task. Because changes cannot drift from the declared configuration once the system boots, upgrades and verification become simpler. Admin teams focus their attention on the Kubernetes cluster and the workloads running on it, instead of on two separate layers with their own risks.

Fairbanks builds and maintains open source private cloud infrastructure based on OpenStack, Kubernetes, Ceph, and related technologies. Organizations struggling with upgrades, patching, or the move to containerization reach out to our team. We think along with you toward a stable, predictable, and secure infrastructure, tailored to your own data center.