# Containers and Orchestration

URL: https://donedeploy.com/devops-services/containers-and-orchestration/

k8s · prod-cluster

node-a

node-b

node-c

**Rollout checkout-api**v2.14, pods replaced one at a time

running

✓

**Autoscaling**3 to 12 nodes, 6 now

6/12

*⎈*

**Kubernetes**EKS · GKE · AKS · bare metal

*AI*

**Runbook suggested**for the restart

How we work

## From design to implementation, and everything after

**Design**the service, its architecture and costs

**Build**infrastructure and code, reviewed

**Run**deployed, secured and watched

**Automate**pipelines, runbooks and AI

**Improve**cost, speed and reliability, every month

The problem

## What goes wrong without it

1

### Works on my machine

Applications behave differently on laptops, test servers and production because each is set up differently.

2

### A cluster nobody dares to upgrade

Kubernetes versions and add-ons fall behind because upgrades were never planned or tested.

3

### Resources wasted or starved

Without sensible requests, limits and autoscaling, some services crash while nodes sit half empty.

### What we do

- Write Dockerfiles that build small, reproducible images
- Choose between managed Kubernetes and Amazon ECS based on your needs, and set it up as code
- Package applications with Helm and deploy them with Argo CD
- Use AI to draft manifests and Helm values and to review changes for missing limits, probes and security settings
- Set up autoscaling, ingress with nginx and secrets management
- Plan and test cluster upgrades so they are routine

### What you get

- Dockerfiles, Helm charts and cluster configuration in your Git
- A cluster or ECS setup defined as code
- A deployment guide for your developers
- Upgrade and incident runbooks

How it works

## From first call to continuous improvement

### Free first consultation

We learn how your applications are built and deployed today and what you need to run.

### Assess

We review your services and recommend Kubernetes, ECS or something simpler, with the reasons.

### Fixed-price proposal

We agree the platform, the services to move and the price.

### Build and move

Engineers containerise the services and move them in stages, with AI reviewing each change.

### Improve after launch

We tune resources, plan upgrades and review cost and reliability over time.

Technologies

## What we use

[!Docker](https://donedeploy.com/technologies/docker/)[!Kubernetes](https://donedeploy.com/technologies/kubernetes/)[!Helm](https://donedeploy.com/technologies/helm/)[!Argo CD](https://donedeploy.com/technologies/argo-cd/)[!Amazon ECS](https://donedeploy.com/technologies/amazon-ecs/)[!NGINX](https://donedeploy.com/technologies/nginx/)[!Prometheus](https://donedeploy.com/technologies/prometheus/)

From the blog

## Related articles

[!AI · November 2, 2026CI/CD for Models and PromptsEvaluation gates, model registries, canary releases and rollback: applying CI/CD discipline to prompts and models in LLM applications.](https://donedeploy.com/cicd-for-llms/)[!AI · September 7, 2026GPU Nodes on Kubernetes Without Wasting MoneyNode pools, MIG, time-slicing, scale to zero and spot instances: practical ways to keep Kubernetes GPU costs under control for AI workloads.](https://donedeploy.com/gpu-nodes-kubernetes/)[!AI · August 27, 2026vLLM vs Ollama vs TGI: Choosing an Inference ServerHow vLLM, Ollama and Hugging Face TGI differ in throughput, hardware support and operations, and which one fits your workload.](https://donedeploy.com/vllm-ollama-tgi/)

Related services

## Often done together

[**CI/CD and GitOps**Releases that are routine, reviewed and easy to roll back*Learn more »*](https://donedeploy.com/devops-services/ci-cd/)[**AI observability**Know what is wrong, and why, before your customers tell you*Learn more »*](https://donedeploy.com/devops-services/monitoring/)[**Cloud and infrastructure**A cloud setup on AWS, Google Cloud or Azure that your team can run*Learn more »*](https://donedeploy.com/devops-services/cloud-providers/)

Questions

## Frequently asked

Do we need Kubernetes?

Not always. For a handful of services, a managed container service such as Amazon ECS can be simpler and cheaper; we recommend Kubernetes when its flexibility pays for its complexity.

Can you take over an existing cluster?

Yes. We review it, bring its configuration under code and fix the risky parts first.

Who runs the cluster afterwards?

Your team, us, or both. We hand over runbooks either way and can operate it under a support package.

Tell us what you run and what you want to change. We reply personally, and the first consultation is free.

[Let's talk »](https://donedeploy.com/lets-talk/)