For some time now, we have been increasingly aware that our customers have a high demand for professional exchange on the subject of Kubernetes. In addition to the individual exchange with our customers and interested parties and the sharing of best practice, we have now also started to deal with various formats such as white papers and academies in order to meet the increasing demand for information.
Kubernetes is only representative for a paradigm of software development, which is to reduce and split up the architecture of large software projects in order to develop and release functionalities faster, to scale a large project over several developer teams and to automate the operation of software projects. Topics such as Continuous Integration and Continuous Development (CI/CD) are becoming increasingly important, and many companies are intensively involved in these areas.
In this article we would like to give you a short insight into the world of container orchestration and especially focus on Kubernetes. In addition to this article we offer you further materials, such as our whitepaper on “Managed Kubernetes”.
How do you actually come up with Kubernetes?
Increasingly we learn from conversations with our customers that container orchestration using Kubernetes is only a later step in a chain of “transformations of software development”. Often the development teams start with solutions that allow to manage individual containers comparatively intuitively. Often we meet Docker Swarm, followed by Hashicorps Nomad.
Only in the course of time, perhaps even with a further growing container landscape, do IT experts start to ask themselves questions about the economics of operating container landscapes. Here, economy stands less for direct costs than for the fact that IT experts try to better and better reflect the growing requirements of the development teams in order to offer the highest possible “developer experience”. The automation of the entire CI/CD landscape of companies is of great importance in this context.
How do you get to your Kubernetes now?
With Kubernetes, a technology has established itself on the market over the last few years that promises to solve some central challenges in the orchestration of containers. Several possibilities have emerged.
These can be divided into the following scenarios: Either companies build up their own expertise in the operation of Kubernetes systems. They make decisions about the right load balancer, about persistent storage and its connection as well as about the right schedulers within Kubernetes. Or alternatively, companies use one of the more or less turnkey solutions offered in the cloud and hosting market.
configure, make K8s, make install
The most obvious approach is first to set up the Kubernetes platform itself. This has the advantage that you get to know the system and technology very well. This way you and your colleagues gain important knowledge and experience in handling and building the container architecture.
You will have to make different decisions. About the right load balancer to route the container services or how you want to deal with persistent storage. In our experience, this approach has proven itself especially for smaller test environments. A good introduction to this topic is the tutorial “Kubernetes the hard way” by Google’s Kelsey Hightower.
Reseller or Managed Hosting Provider
If do-it-yourself construction is out of the question for you, the search for a suitable service provider would be a good option. There are different approaches here. You can simply hand over the complete platform to a managed hosting provider for operation. This means that you are handing over the reins, but in return you have a service provider at your side whom you can engage for certain services and functionalities.
The disadvantage of Managed Hosting Providers, in our opinion, is that there is comparatively little standardisation in the area of container orchestration. The providers have different solutions and products available for identical requirements. Some managed hosting providers also act as resellers of the large Hyperscalers and then leave you the choice between a, mostly purchased, “framework approach” and the solution modules of the resold Hyperscalers.
As a manufacturer of our own solution, we know how important uncompromisingly high expertise in complex container orchestrations such as Kubernetes is for the success of projects that rely on this framework. We are therefore comparatively sceptical about offers from service providers who do not create their own added value.
Cloud provider plus modern GUIs
The Hamburg-based company loodse, for example, offers a very interesting and different approach. Loodse tries to reduce the complexity of a container orchestration at least on the GUI level and to simplify the management of workloads of any kind with a modern web interface. This software can control both the APIs of the Hyperscaler and the framework approach mentioned above.
This solution mainly addresses problems on the organizational and administrative level, but is not a stand-alone Kubernetes solution in itself, but must be combined with other services. Nevertheless, from our point of view this is an excellent approach to successfully provide container orchestration based on Kubernetes.
Native cloud provider
In contrast to the variants presented so far, each of which consists of different tools, services and offers to map a complete container orchestration, there is of course still the “Cloud Native” approach. Here, the respective providers rely on a highly automated complete solution, which can often be extended by specific services.
With “turnkey” Kubernetes solutions, the provider takes care of continuous development and updating, operation, capacity management, platform monitoring and security updates.
However, even “turnkey” solutions may still throw a few stones in your path. For example, you may still have to decide on a load balancer or persistent storage, which in the best case can then be integrated quite easily via the vendor’s platform services.
A disadvantage of working with large cloud native providers is often the less user-friendly interfaces. Unfortunately, you can expect little support or exchange here, as the companies usually focus completely on a self-service approach.
This is exactly where the native cloud offer of gridscale stands out. You have access to a user-friendly and intuitive interface that already provides all important pre-configurations for your individual workload. This allows you to get started with container orchestration in a few minutes and at the same time you can customize it later if your platform requirements change.
Well, and now
This article should give you a small overview. With this information you will be able to make a first decision. If you want to get deeper into the topic of Kubernetes, then download our whitepaper about Managed Kubernetes on gridscale.
Henrik Hasenkamp | CEO - Strategie & Partner
Henrik is gridscale’s CEO. In this function, he is primarily responsible for the strategic development of gridscale, as well as important partnerships. Henrik has experience of 15 years in the hosting industry. In his last role, he took charge of the international realignment of the dedicated server business of one Europe’s largest hosting companies.