Daily Exercise 001: How-Kubernetes-Works

Day 1 Script

Kubernetes is fundamentally about taking a bunch of virtual or physical machines and transforming them into a unified API surface. This allows developers to interact with containers and orchestrate their applications without thinking about the underlying machines. We often abbreviate Kubernetes as k8s because there are eight letters between the K and the S. The Kubernetes API presents several core pieces, including pods, services, and deployments. A pod is a collection of containers co-located on a single machine. A service acts as a load balancer to bring traffic to a collection of pods. Finally, a deployment uses a replica set to replicate a container multiple times for better availability and scale. Users generally interact with this JSON based API through the kube control command line tool or other similar tools.

Day 1 Translate

Day 2 Script

Let's look at how we put these pieces together to build a real application. Suppose I have a simple web app packaged as a container image. To run this application on the Kubernetes API, I need to create pods. I do this by creating a deployment, which is typically represented as a YAML file. For example, I might create a file named deploy yaml and use the kube control apply command to send it to the API server. If I specify three replicas in that file, the API server will create three pods. The scheduler then places these three containers onto the virtual machines that act as hosts for the service. This is just the beginning, as the deployment object manages and ensures that the requested number of replicas is always running.

Day 2 Translate

Day 3 Script

Once the application is up and running, I need to expose it so that users can consume it. To do this, I create a service. A service is a load balancer that knows how to take traffic from the outside world or from another service inside the cluster. It then load balances that traffic down to the containers. If I want, I can attach this service to an external load balancer. When I do this, Kubernetes is responsible for talking to the cloud and requesting that a load balancer be created. This process generates an IP address that is mapped to the cloud load balancer. This IP address is then connected to the service and all of its associated traffic.

Day 3 Translate

Day 4 Script

When an end user accesses the application, they talk to the external IP address. The traffic flows through the cloud load balancer to the service, which then distributes it to all the containers. This setup highlights the declarative and self healing nature of Kubernetes. For example, if I change the number of replicas in the deployment from three to four, Kubernetes will notice the discrepancy. It will realize only three replicas exist and automatically create a fourth one. Traffic will then start flowing from the load balancer to this new replica. The end user does not even notice that the application has been scaled up. This reliability is one of the main reasons why Kubernetes is so effective for modern applications.

Day 4 Translate

Day 5 Script

Kubernetes provides a great way to deploy reliable and scalable applications without affecting end users. In addition to scaling, you can also perform a rolling deployment. This allows you to move from version one of your container to version two in place. During this process, the end user will not notice any interruption at all. The system manages the transition smoothly by replacing old containers with new ones gradually. This ensures that the application remains available throughout the update process. We will dive deeper into the technical details of how rolling deployments work in the next episode. For now, it is important to understand how these core components work together to provide a seamless experience for both developers and users.

Day 5 Translate