There is a category of problem that only appears after deployment, and it is invisible to a development team that has never carried a pager. A rollout that drops requests because pods are removed from service before they stop receiving traffic. A disruption budget expressed in a way that lets a node drain take the last replica. A scheduled job running on three replicas at once because nothing makes it a singleton. These are not advanced topics. They are the difference between a system that is up and a system that appears up.
What the work involves
- Cluster and namespace design, or a candid assessment that you do not need Kubernetes at all
- Deployment manifests with readiness gates, pre-stop hooks and minimum-ready periods as standard, not as an afterthought
- Disruption budgets expressed so a node drain cannot take your last replica
- Pipelines that build, guard and deploy, with the guards able to stop a release
- Metrics, logs and traces wired in before launch
- Runbooks written for whoever is on call, including the case where that is not us
Why we are credible at this
We run an RKE2 cluster carrying 106 deployments and 267 pods across 19 namespaces, with our own storage, ingress, certificate automation and virtualisation. The deployment safety rules we apply to client work are the ones we wrote down after breaking our own.
What moves the price
We quote after discovery and the number is fixed before anything starts. These are the things that move it, so you can see roughly where your project sits before you talk to us.
- Whether you are starting fresh or inheriting infrastructure that has drifted from its configuration
- Compliance and data-residency requirements, which constrain where things may run
- How many environments need to exist and stay consistent with each other
- Whether your team needs to operate it afterwards, which makes documentation and training part of the work
Proof we have done this
Questions we get asked
Other App Development work
Tell us what you are trying to do
A short conversation is usually enough to tell whether we are the right people. If we are not, we will say so and point you somewhere better.