Lightweight Kubernetes: Evaluating K8s vs. K3s for Your Project

The Kubernetes orchestration tool has been top of mind for development teams since its release in 2014. Kubernetes, or K8s, is now the go-to tool for launching and managing containers in cloud environments. It’s an open source, stable, versatile platform that scales to support large clusters in sizeable, and diverse workloads. Major cloud platforms and enterprise OS environments also support Kubernetes.

With the number of Kubernetes use cases and architecture designs, you might assume  everyone would use the tool. But it can be challenging to find the right project to take advantage of Kubernetes, and depending on resources, it can be complicated to implement and manage once up and running. 

If you’re interested in using Kubernetes–without the complexity and increase in resource usage–there is a light and easy container orchestration solution called K3s. Get it, K3s?

Up Close

K3s is part of a new generation of lightweight container orchestration tools. The project developers at Rancher describe K3s as a solution great for “situations where a Ph.D. in K8s clusterology is infeasible.” Similar solutions include Minikube, which is currently in development, and Canonical’s MicroK8s, which is easy on resource consumption but not as easy to configure and use as the other lightweight options. 

K3s and others are not exactly alternatives to K8s. However, they are scaled-down Kubernetes distributions, with much of the same code at the core but using half of the memory needed than K8s. To achieve this smaller size, the developers reduced the number of dependencies and removed some of the add-ons. The goal is to provide the core functionality of Kubernetes without the bloat—and in a form that is easier to configure and use.

Initially designed for edge computing and IoT settings, K3s works equally well in simple cloud computing deployments

K3s or K8s?

The choice between K3s or K8s depends on the use case. In general, if you are expecting a high-volume scenario with a number of applications across a large distribution of clusters, K8s is the best option. But if you’re looking for something to support a smaller, lighter workload that will spin up quickly and won’t experience heavy data loads, K3s could be the better choice. Whether you’re exploring containerization or starting a smaller project, your team’s people power and individual experience with Kubernetes may point you toward K3s based on ease of use and set-up time.

It’s also important to note that K3s doesn’t include a distributed database, which makes it unsuitable for distributed clusters sharing a data store. Rancher maintains a distributed version of the SQLite database called Dqlite, which can be used to add distributed database capabilities to K3s. Implementing this kind of solution can add time to configuration and orchestration processes.

Another important difference is that K3s runs as a single binary, whereas K8s operates various components as separate processes. The more sophisticated architecture of K8s improves stability and scales more efficiently to high volumes, making it significantly more complicated.

In general, the benefits of K3s are also its shortcomings if you apply it to a project that isn’t classically successful on Kubernetes. For instance, one advantage of K3s is its small footprint, it  reduces the number of drivers and extensions. But if your work requires maximum flexibility and supports a wide range of hardware and software components. These limitations will come at a cost in the form of customization and reconfiguration.

Ultimately, K3s or K8s are great options to containerize and orchestrate your workloads and help you manage clusters. If your cloud environment is just getting started, looking for early proof-of-concept, or just need a lighter approach, check out K3s first.