Docker + portainer vs k8. EILI5

K8s is short for Kubernetes, it's a container orchestration platform. It's a complex system but the basic idea is that you can run containers on multiple machines (nodes). This can help with scaling out applications and achieving High Availability (HA).

Current versions of K8s do not require docker to run, but nevertheless you can run OCI containers (sometimes referred to as docker containers).

Let me know if you have any questions.

I've been working on a wiki to provide guides and resources on setting up your own K3s cluster, which is a flavor of Kubernetes that comes with a little less bloat by Rancher.

Docker is more like the lower level when it comes to container management. It's really good in development, but lacks lots of features and structure for production. It becomes harder to maintain an entire set of Docker containers and remote management isn't as easy either. Which is what Portainer is for but you often might end up accessing via SSH anyway.

Kubernetes builds on top of containers, including Docker as a possible runtime, to add all the tools to schedule and manage containers and workloads. Kubernetes has a couple of resource types you should be aware of, that help you deploy in a more scaled manner:

• Namespaces: Organizational units you can use to separate your deployments. I have each deployment in its own namespace, so even if I named two pods the same, they can't clash with one another.

• Pods: The smallest container unit in Kubernetes you can deploy. A pod can contain multiple containers, though, but these are akin to Docker/Docker Compose containers.

• Deployment/StatefulSet/DaemonSet: These are Kubernetes workloads that handle the way they scale (replicate) pods in different ways. A Deployment simply creates replicas, while a StatefulSet assigns a new PVC to each pod, and DaemonSet ensures that a certain number of pods are running on each node.

• Secret/ConfigMap: Using Secrets and ConfigMaps, you can add data, such as environment variables or configuration files to the cluster, and use them in your containers by mounting them as files or as environment variables.

• Services: Since Kubernetes is built for scaling, you'll often be running more than one pod for a singular application. Services expose those pods' ports and take care of load-balancing for you.

An important aspect of Kubernetes is how it handles scaling. It's designed with multiple nodes in mind, and usually containers shouldn't care what node they're running on. Most modern apps that use block storage and external databases, or if you have distributed storage such as Longhorn, can be replicated multiple times or even a certain number of times per node, to spread the load between them.

Kubernetes also brings with it a massive ecosystem. A lot of the work has already been done in the Kubernetes space for you, so there are Kubernetes-native automation tools like ArgoCD and Flux, that monitor changes in Git repositories for your manifests (similar to docker-compose.yml files) and automatically apply them.

There are also lots of controllers and operators for typical things you might want to do, such as Nginx with Cert Manager and Traefik, which is built-in with K3s.

Let me know if this helps and if you need any other advice before making the decision!

You could also have a look at Nomad. It’s much more simple compared to k8s