Many services, one platform
A product made of several APIs, workers and front-ends that need consistent deployment, networking and configuration.
Technologies · Kubernetes
Hire Kubernetes engineers who can tell you when not to use it
Kubernetes is powerful and expensive to operate badly. Our engineers write the manifests, charts and pipelines, and keep the cluster understandable for the people who inherit it.
What is Kubernetes?
When you hire Kubernetes engineers, you get people who build and run applications on Kubernetes, the open-source system that schedules containers across a cluster of machines, restarts them when they fail, and scales them with demand. Teams describe the desired state of each application in declarative manifests, and Kubernetes continuously works to match it. The engineering work covers cluster setup, packaging with Helm, GitOps delivery, networking, secrets, autoscaling and observability.
When Kubernetes is worth the complexity
Variable traffic
Workloads with daily or seasonal peaks where pods and nodes should scale out and back in automatically.
High availability across machines
Applications that must keep serving when a node or availability zone fails, with pods rescheduled onto healthy nodes.
Portability between clouds
Organizations that want the same deployment model on EKS, GKE, AKS or their own hardware.
When we recommend something else
- You probably don't need Kubernetes for one or two applications on a single server — Docker Compose with good backups is cheaper to run and easier to understand.
- A single Odoo instance for one company rarely benefits; a well-sized server or Odoo.sh is simpler to operate.
- No one on the team can own the cluster — upgrades, certificates and security patches arrive on a fixed schedule whether you are ready or not.
- A managed platform such as Cloud Run, App Runner or a PaaS already runs your containers with less operational work.
What our Kubernetes engineers deliver
Cluster foundations
Managed clusters on EKS, GKE or AKS provisioned with Terraform, node pools, namespaces, RBAC and network policies.
Helm charts and manifests
Charts with sensible defaults, resource requests and limits, readiness and liveness probes, and pod disruption budgets.
GitOps delivery
Argo CD or Flux syncing the cluster from Git, so every change is reviewed and can be rolled back by reverting a commit.
Platform add-ons
Ingress or Gateway API routing, cert-manager for TLS, External Secrets for credentials, autoscalers, and Prometheus and Grafana monitoring.
A production cluster, layer by layer
L1
Traffic entry
A cloud load balancer in front of an ingress controller or Gateway API implementation, with TLS certificates issued by cert-manager.
L2
Workloads
Deployments, StatefulSets and Jobs running containers with resource limits, probes and horizontal pod autoscaling.
L3
Configuration and secrets
ConfigMaps for settings and Secrets synced from a vault or cloud secret manager, never committed in plain text.
L4
Delivery
CI builds and scans images; Argo CD or Flux applies the Helm releases recorded in Git.
L5
Cluster and nodes
Control plane and node pools, scaled by Cluster Autoscaler or Karpenter, separated by namespaces and network policies.
L6
Observability
Prometheus metrics, Grafana dashboards, centralized logs and alerts on the signals users actually notice.
Ecosystem we work with
- Helm
- Argo CD
- Flux
- cert-manager
- Gateway API
- Ingress controllers
- External Secrets Operator
- Karpenter
- Prometheus
- Grafana
- Kustomize
- k3s
What a well-run cluster gives you
Self-healing workloads
Failed containers restart and pods move off unhealthy nodes without someone logging in at night.
Capacity that follows demand
Pod and node autoscaling add capacity at peaks and release it afterwards, if requests and limits are set realistically.
Auditable changes
With GitOps, the state of the cluster is described in Git, so every change has an author, a review and an undo.
Zero-downtime releases
Rolling updates with readiness probes send traffic to new pods only once they are ready.
Managed or self-managed Kubernetes?
| Consideration | Managed (EKS, GKE, AKS) | Self-managed (kubeadm, k3s, RKE2) |
|---|---|---|
| Control plane | Run and patched by the cloud provider | Run, backed up and upgraded by your team |
| Cost model | Control-plane fee plus nodes | Hardware or VMs plus engineering time |
| Cloud integration | Native load balancers, storage and IAM | Configured separately, e.g. MetalLB and CSI drivers |
| Data location | Provider regions | Any data center, including on-premises |
| Good fit | Most teams running in a public cloud | On-premises, edge or strict data-residency requirements |
Related services, technologies and industries
Sources and further reading
Official documentation and guidance if you want to read more about this topic.
Kubernetes questions
What does a Kubernetes engineer do?
A Kubernetes engineer designs, deploys and operates applications on Kubernetes clusters. Day to day that means writing manifests and Helm charts, setting up GitOps delivery, managing networking, certificates and secrets, tuning autoscaling, upgrading clusters and responding to alerts.
Do we need Kubernetes?
Probably not if you run one or two applications on a single server. Kubernetes pays off when you have many services, need failover across machines, or face traffic that varies enough to require autoscaling, and when someone can own the cluster long-term.
How is this different from your Kubernetes consulting service?
This page describes Kubernetes as a technology and the engineering skills we apply to it. Our Kubernetes consulting service is the structured engagement for assessing, designing or migrating to a cluster.
Which managed Kubernetes service should we choose?
Usually the one from the cloud you already use. EKS, GKE and AKS all run conformant Kubernetes; the practical differences are in pricing, integration with each provider's networking and identity, and how upgrades are handled.
Can Odoo run on Kubernetes?
Yes. Odoo can run as a Deployment with a shared filestore on persistent storage, a managed PostgreSQL database, and a separate route for the real-time websocket worker (long-polling in versions before Odoo 16). It makes sense for hosting many Odoo instances, but is usually more than a single company's ERP needs.
What is GitOps in Kubernetes?
GitOps means the desired state of the cluster is stored in a Git repository and a controller such as Argo CD or Flux continuously applies it. Changes are made by pull request, and rolling back is reverting a commit.
Next step
Weighing Kubernetes, or short of people to run it?
Tell us how many services you run, where they're hosted and who looks after them today. We'll tell you whether Kubernetes fits and what it would take to operate it well.