What is Kubernetes and why is it so attractive?

Kubernetes (or 'k8s' in techie talk) is a system for managing large groups of containers and containerised applications. This technique allows you to control the operation of containers – a process called 'Container Orchestration'. Kubernetes was originally developed by Google, but it eventually found a place in the open source community. Thanks to an open standard such as Docker, containers have become quite popular. With Kubernetes, you can manage a lot of containers – really a lot – without losing the big picture.

Today, scalability is achieved by horizontal scaling: more servers, instead of larger and more powerful ones. Microservices, which are an important trend in application development, makes us partition large software blocks into multiple smaller ones. This means you have a lot more to manage: tens, hundreds or even thousands of nodes. In a typical cloud approach, these are no longer 'servers' but 'containers'. Fortunately, Kubernetes makes it possible to manage all this in a different way.

Check out our managed Kubernetes solutions

The orchestra and the sheet music – How does Kubernetes work?

Kubernetes enables orchestration, i.e. automation of container management. What makes containers so special is the fact that they contain an autonomous piece of code, which is often impossible to change. They can be started within seconds, either manually or automatically. Kubernetes starts the containers, takes care of things like load balancing, failover, routing of network traffic in case of high traffic, easy (automatic) scaling, but also ensures downscaling of resources when things are quiet. If a container crashes, Kubernetes steps in and starts a new one.

And that is exactly what makes Kubernetes so special – you can have containers start and stop fully automatically, based on predefined criteria. Examples of such criteria include the following:

• A minimum number of instances that must be enabled for each container
• The response times for each of these instances must remain below your set thresholds
• When your monitoring system detects a problem, you can immediately add an additional instance
• And you can automatically divert incoming traffic away from the suspected problem and restart the instance that crashed.

These are just a few examples of orchestration rules you can set using Kubernetes. The key idea being to anticipate future issues. Although you can never foresee all possible obstacles, most problematic situations are covered in advance by a scenario. That is the foundation for advanced automation. You can actually solve the problems before they even occur.

This, however, requires a completely different mindset. Whereas you used to be a virtuoso on your favourite instrument as a system administrator, now you are the conductor who ensures the ensemble is in perfect harmony. And the basic rules that govern the ensemble form your sheet music.

Different game rules – What makes Kubernetes so special?

The great advantage of Kubernetes is that you eliminate the human factor, which is the weakest link. Deploying containers within an infrastructure cluster, as well as starting and stopping additional nodes, can in principle be done manually. But doing so means that you will not be able to react as quickly as if those processes were running fully automatically.

The essence of clustering with containers is precisely to minimise the impact of a possible system error by making sure that all components are redundant and that resources can be scaled up elastically. With Kubernetes, this happens immediately, even before you had time to take action as a human administrator.

Orchestration with Kubernetes also allows you to fully automate the decision to provide extra nodes, and to divert traffic away from any possible problem. A master server monitors the cluster infrastructure, which starts and stops additional nodes based on the rules that have been set. The nodes are virtually isolated pieces of infrastructure on which each container runs, including the network settings for access and load balancing. Containers that interoperate and are highly interdependent can also be grouped into pods.

The management of your infrastructure and application components is thus much less 'hands-on'. Because you manage the master server instead of managing each node individually. Knowledge of formats such as JSON (JavaScript Object Notation), REST (Representational State Transfer) or YAML is not a luxury when it comes to laying down the rules of the game. From a conceptual point of view, we are a million miles away from 'system administration' in the classic sense of the term.

Flexible and flawless – What are the benefits of using Kubernetes?

Are there any drawbacks to using Kubernetes?

Combell & Kubernetes

Do you find Kubernetes an interesting option, but lack the technical expertise required to configure and manage it? No problem! Combell offers you Managed Kubernetes solutions, in order make your life easier or even to fully relieve you of your worries. This way, you will still benefit from the necessary guarantees in terms of security, reliability and support.

Scaling up quickly and flexibly obviously only works as long as the physical infrastructure makes it possible. With containers, you can use the available resources very efficiently, with minimal overhead. Nevertheless, the physical infrastructure must be able to cope with peaks, or to scale up in no time. Combell provides affordable packages that allow you to grow rapidly or to develop in line with your actual needs.

Are you interested? Or do you have any questions?

Feel free to contact our specialists