Explaining the Basic Concepts of Kubernetes

One of the critical challenges Kubernetes addresses is the efficient management of application workloads.

Many organizations typically have a wide range of applications running on their production servers- either bare metal or virtual machines. These applications are often tightly coupled to the underlying infrastructure and are not portable across different environments (for example, a software installation on a specifically configured Linux server). That makes it difficult to move applications from one environment to another, such as from a test environment to a production environment or from a failed server to a healthy one.

Using containers for shipping and operating applications can help you towards a more portable and efficient way of managing applications. Containers package applications with all their dependencies as a portable and executable unit (container image). The application can be shipped and run on almost any environment with an installed container run time/engine.

Also, from an economic point of view, most servers are often over-provisioned. Consequently, companies pay a fixed price for permanently allocated capacities (for example, computing power, storage capacity), which are not fully utilized. This is also commonly referred to as idle costs / unused capacity costs. This is particularly true for servers operating workloads that do not run at full capacity at all times (for example, during nights or weekends).

Especially, if you want to operate hundreds or even thousands of containers in production simultaneously, you definitely need a way to manage them efficiently.

However, efficiently managing workloads across multiple hosts in your IT landscape can be challenging and complex.

That is where Kubernetes comes in. Kubernetes can manage and orchestrate workloads in a highly efficient way. It can do so by automatically scheduling workloads to the correct host, scaling them up and down, and rolling out new versions of your applications.

For example, based on the current high demand, Kubernetes can automatically scale applications horizontally (that is, by adding more instances) and balance incoming traffic across those instances. Kubernetes can also automatically move applications from one host to another if the current host becomes unavailable or if the host is over-provisioned. Kubernetes plays Tetris with containers to make sure that applications are always running and that they are running in the most efficient way possible.

For these reasons, Kubernetes is also known as Container Orchestrator.

Kubernetes has several features that allow you to manage containerized applications in a scalable and reliable way. Among the most important features are the following:

Service discovery and load balancing

Containers can be exposed via their DNS name or using their IP address. Adopting a load balancing approach, Kubernetes can distribute network traffic to stabilize deployment when container traffic is too high.

Storage orchestration

Storage can be automatically integrated with Kubernetes using a storage system of your choice, such as on-premises storage, public cloud providers, or a networked storage system.

Automated rollouts and rollbacks

Kubernetes allows you to update containers using rolling updates. Kubernetes starts new containers and then kills old ones. This feature also ensures zero downtime during the deployment. Should anything go wrong during the rollout process, Kubernetes rolls back the change for you.

Immutability and self-healing

Everything in Kubernetes is defined declaratively. You specify the desired state of an application or infrastructure, and Kubernetes always ensures that the current state matches the desired state. When a container crashes, Kubernetes restarts it for you. And when an entire node goes down, Kubernetes moves the containers to a healthy node and shuts down containers that don't respond to your custom health checks.

Secret and configuration management

You can use Kubernetes to store and manage confidential information, such as passwords, tokens, and SSH keys. You can deploy and update secrets along with application configurations without rebuilding your container images or revealing sensitive information in applications.

Extensibility and Ecosystem

The Kubernetes API comes with a standard set of implementations and objects. Kubernetes often defines the API but not the concrete implementation. This allows you to use and switch between different implementations of the same API. For example, you can use different storage systems for persistent volumes. You can also use different container runtimes, such as Docker, rkt, and Rocket. Over time, the Kubernetes ecosystem has grown to a size where you can find a solution for almost any problem. For example, there are solutions for logging, monitoring, service meshes, ingress, and many more. We will cover some of them later in this course.

To get a full list of all Kubernetes features, check out the official documentation.

For further reading, see the following sites:

• Kubernetes
• Cloud Native Computing Foundation (CNCF)
• CNCF Members