Solutions
Platform Upgrades
Keep your platform current without turning each upgrade into a high-stakes event. We audit your current versions and deprecated API usage, sequence upgrades to minimize blast radius, and build the automation and runbooks to make version increments routine rather than exceptional.
The Business Problem
Falling behind on platform versions — upgrade debt that makes each upgrade harder and riskier than the last
The Challenge
Kubernetes releases a new minor version approximately every four months, and each version is supported for roughly fourteen months. That sounds manageable — until it isn’t. Organizations that let their cluster version drift even two or three releases behind find themselves in increasingly dangerous territory: security patches unavailable, deprecated APIs in use, and a large upgrade delta that requires careful, coordinated testing.
The same pattern plays out with node operating systems, container runtimes, cloud provider managed services, and the platform software stack (ingress controllers, cert-manager, observability tooling). Individual components have their own release cycles, and keeping everything compatible requires active attention.
Our Approach
We treat platform currency as an ongoing operational discipline, not a one-time project. The goal is to make upgrades routine — small increments done frequently — rather than high-stakes major events done under pressure.
For organizations that have already accumulated upgrade debt, we start with a current-state audit: what versions are running, what deprecated APIs are in use, what workloads are at risk during an upgrade. We design a sequenced upgrade plan that minimizes blast radius and validates each step before proceeding.
We implement upgrade automation where it makes sense — automated testing of upgrade paths in staging environments, with clear promotion criteria before production. For managed Kubernetes services (EKS, GKE, AKS), we configure upgrade channels and maintenance windows aligned with your change management process.
Technology Options
- kube-no-trouble (kubent) — scans clusters for deprecated API usage before upgrades, identifying what needs to change
- Pluto — similar API deprecation detection, with CI integration for catching issues in manifests and Helm charts
- cluster-api (CAPI) — declarative Kubernetes cluster lifecycle management, enabling version upgrades via infrastructure-as-code
- Managed Kubernetes upgrade channels — EKS managed node groups, GKE release channels (Rapid / Regular / Stable), AKS auto-upgrade channels
- Helm chart CI — automated testing of Helm charts against new Kubernetes API versions using chart-testing and kind clusters
- Upgrade runbooks — documented, version-controlled procedures tested in non-production before production rollout