Table of Contents
1. The Cost Savings of On-Prem Kubernetes
2. Get More Control by Running Kubernetes On-Prem
3. Leverage On-Prem Bare-Metal Servers to Enhance Performance
4. Minimize Lock-in Risks by Staying On-Prem
Kubernetes is a cloud-native technology, but that doesn't mean it always has to run in the cloud. On the contrary, there are a variety of reasons why you might choose to run Kubernetes on-premises, rather than hosting it on VM instances running in a public cloud or via a cloud-based managed Kubernetes service.
This article unpacks four advantages of on-prem Kubernetes. As you'll learn, Kubernetes on-prem doesn't make sense for every organization or use case, but it does beat cloud-based Kubernetes in some scenarios.
1. The Cost Savings of On-Prem Kubernetes
Probably the single most important benefit of on-prem Kubernetes is the potential for lower cost.
To be sure, on-prem Kubernetes is not necessarily more cost-effective than cloud-based Kubernetes. But it can be, especially if you already own sufficient on-prem infrastructure to host your clusters and the infrastructure is a good fit for your workload requirements.
In addition, with on-prem Kubernetes, you don't have to pay costs like data egress fees or control plane management fees, which apply to most cloud-based Kubernetes environments.
Cloud-based Kubernetes may make more sense from a cost perspective if your workloads are highly scalable or unpredictable, in which case the ability to take advantage of pay-as-you-go pricing for scalable cloud infrastructure is advantageous. But in other situations, on-prem Kubernetes may deliver a lower total cost of ownership.
2. Get More Control by Running Kubernetes On-Prem
In almost all cases, on-premises Kubernetes clusters offer the advantage of more control. When you own and manage the entire infrastructure stack that hosts your clusters, you can configure compute resources, networking, storage, and so on in whichever ways you want.
In the cloud, configuration options are much more restrictive. You may be able to pick and choose from different VM instances to run your nodes, and you'll have a fair amount of control over how Kubernetes itself is configured. But you'll be limited to whichever networking and storage options your cloud provider supports.
3. Leverage On-Prem Bare-Metal Servers to Enhance Performance
Running Kubernetes on-prem makes it easy to take advantage of bare-metal servers to improve the performance of workloads. With bare metal, you don't waste any infrastructure resources on hardware virtualization, so more resources are available to your workloads. Bare metal also makes it possible for some workloads, like GPU-accelerated containers, to take advantage of special hardware features that typically wouldn't be accessible from within virtual machines.
Not all Kubernetes workloads benefit from bare-metal hosting, but for those that do, on-prem is the way to go. Bare-metal server instances are available in some clouds, but they are expensive, and they don't provide as many device or configuration options as you'll get from running Kubernetes on your own on-prem hardware.
4. Minimize Lock-in Risks by Staying On-Prem
Not all cloud-based Kubernetes environments lock you into a particular vendor's ecosystem or platform. But some do, especially if you choose a managed Kubernetes service. Migrating from, say, Amazon EKS to Azure AKS is not trivial, and so you may end up locked into a particular cloud if you opt for cloud-based managed Kubernetes.
When you run Kubernetes on-prem, your lock-in risk is lower. You're more likely to end up with a generic Kubernetes configuration that could be ported to any infrastructure environment. You may still be tied to a certain distribution, but it's easier to move from on-prem Rancher or OpenShift to cloud-based Rancher or OpenShift, for example, than it is to move from one cloud-based managed Kubernetes service to another.
Conclusion
Again, on-prem Kubernetes isn't the right choice for every workload or every scenario. But it has its selling points, especially for organizations keen to minimize their Kubernetes costs, maximize their control, optimize performance, and avoid becoming locked into a specific Kubernetes hosting setup.
About the author
Christopher Tozzi is a technology analyst with subject matter expertise in cloud computing, application development, open source software, virtualization, containers and more. He also lectures at a major university in the Albany, New York, area. His book, “For Fun and Profit: A History of the Free and Open Source Software Revolution,” was published by MIT Press.
