Hybrid Kubernetes is a Kubernetes deployment model for running containerized applications seamlessly across on-premise data centers and the public cloud.

Outline

• Why do we need hybrid Kubernetes
• Hybrid Kubernetes architecture
• Hybrid Kubernetes deployment models
• Challenges of hybrid Kubernetes
• Wrapping up

Why do we need hybrid Kubernetes

Running Kubernetes on the public cloud is simpler than in on-premise data centers.

Cloud providers offer managed Kubernetes services to help you get started quickly. A managed Kubernetes service from a cloud provider interworks smoothly with the cloud provider’s other services like load balancers, firewalls, virtual private networks, etc.

You don’t get those amenities in on-premise data centers. Also, the agility of the public cloud is no match for on-premise data centers.

But, not everyone wants to be always on the public cloud.

Hybrid Kubernetes models can fulfill technical and business needs that the public cloud alone cannot meet.

• Keeping sensitive data on-premise

  Data governance and cybersecurity policies demand enterprises to take extra precautions on sensitive data. Such policies could mandate you to store sensitive data in an on-premise data center along with the applications handling those data.   To take advantage of the public cloud for other workloads, you can use a hybrid Kubernetes model where Kubernetes clusters in the on-premise data centers will be running the applications that handle the sensitive data, and the Kubernetes clusters on the cloud will be running the rest of the applications.

• Intermediate step for enterprise cloud migration

  Migrating an on-premise enterprise cloud to the public cloud will likely be a project with an extended time span. With high expectations for business continuity, you cannot risk migrating all your enterprise applications to the cloud in one go.   To reap the benefits of the public cloud from an early stage of the project, you can employ a hybrid Kubernetes architecture and gradually shift the workloads from the on-premise data center to the public cloud.

• Running latency-sensitive workloads closer to the user

  When latency matters performance, running applications closer to the user, can boost the user experience.   A prime example is telecom service providers running cloud-native 5G applications with stringent latency requirements. Cloud gaming is another arena that benefits from running applications closer to the users.   Hybrid Kubernetes can fulfill this requirement effectively. You can run the latency-sensitive applications in regional data centers while running the rest of the applications on the cloud.

• Using the public cloud for disaster recovery

  Disaster recovery is an expensive endeavor in on-premise data centers because you need to build redundant capacity.   Instead of investing in extra capacity in on-premise data centers, you can make use of the public cloud for disaster recovery with hybrid Kubernetes by moving the workloads to the public cloud in the event of a failure.   The public cloud is available on demand. You can provision the cloud resources at the time of the recovery so that you will not be paying for unused capacity.

• Using the public cloud for overflow or burst traffic

  Imagine that you are running your software applications primarily in private data centers. You need to handle two ultra-high seasonal peaks every year. During these peaks, your traffic load goes up to tenfold the traffic on a normal day.   It would be a waste of money to build capacity in on-premise data centers to handle such seasonal peaks. That capacity will remain unutilized for 90% of the time.   A hybrid Kubernetes model can overflow traffic bursts to the public cloud. Since the public cloud is available on demand, you can pay only for what you consume during the peaks.

• Gain more bargaining power

  In any negotiation, your bargaining power goes up when you have alternatives to fulfill your needs.   Hybrid Kubernetes gives you the option to expand capacity in the on-premise data center when needed.   So, you can leverage hybrid Kubernetes to get better pricing from cloud providers.

• Optimize cloud spending

  You can optimize public cloud spending by retaining certain parts of your applications in on-premise data centers.   Running CPU-hungry tasks in the on-premise data center, aggregating data locally before sending it to the cloud, and caching frequently used data in the on-premise data center are some examples where hybrid Kubernetes can help you optimize cloud spending by running a part of the application on-premise.

Hybrid Kubernetes architecture

In hybrid Kubernetes, you run Kubernetes clusters in on-premise data centers and the public cloud. Then, you can flexibly distribute your workloads between these Kubernetes clusters.

Hybrid Kubernetes deployment models

The Kubernetes distribution is a key factor in hybrid Kubernetes. You have three deployment models to choose from based on what Kubernetes distro you choose.

A Kubernetes distribution (aka a distro) is a product or service that bundles Kubernetes with a set of other tools. A Kubernetes distro also can have a layer of features and functions added on top of open-source Kubernetes.

#1 Cloud agnostic hybrid Kubernetes

You can build a cloud-agnostic hybrid Kubernetes model by running a cloud-agnostic Kubernetes distro in both the on-premise data center and the public cloud.

Your cloud-agnostic Kubernetes distro could either be a self-managed Kubernetes or a managed Kubernetes distro from an independent vendor who is not a cloud provider.

If you are already running Kubernetes in an on-premise data center and want to extend your Kubernetes deployment to the cloud for disaster recovery or handling traffic bursts this model will work best for you because you will get an identical experience in both clouds.

Pros

• Avoid vendor lock-in

  You are not depending on the cloud provider’s managed Kubernetes. So you have less dependency on the cloud provider. Should you decide to switch providers or go multi-cloud in the future, you have fewer roadblocks.

• Use the same toolset to manage Kubernetes

  You can use the same toolset to manage your Kubernetes clusters either on-premise or on the cloud.

Cons

• Interworking with the cloud info

  A cloud-agnostic Kubernetes distro needs to interwork with services like virtual private cloud, load balancers, firewalls, etc., in each public cloud platform.   Sometimes there could be limitations in these interworkings when compared with the managed Kubernetes from the cloud provider.

#2 Distinct hybrid Kubernetes

In the distinct hybrid Kubernetes model, you will run different Kubernetes distros in the on-premise data center and the public cloud. You can run any managed or self-managed Kubernetes distro in the on-premise data center and choose cloud provider-managed Kubernetes on the cloud.

This model is suitable for enterprises that choose to run hybrid Kubernetes as an interim solution during cloud migration.

Pros

• Get the full benefit of managed Kubernetes on the cloud

  Since you are using the cloud provider’s Kubernetes service on the cloud, you are guaranteed smooth sailing.

Cons

• Managing different Kubernetes distros

  It would be a bit awkward to operate two different Kubernetes distros that have different tools and interfaces. Also, their features and functions may not be identical.

#3 Cloud provider-managed hybrid Kubernetes

To build a cloud provider-managed hybrid Kubernets model, you need a cloud provider who offers a version of their Kubernets that you can deploy in an on-premise data center. Not all cloud providers support this.

Your clusters in the public cloud will run the cloud provider’s managed Kubernetes and the on-premise clusters will run the on-premise version of the cloud provider’s managed Kubernetes.

This model is suitable if you are already running Kubernetes on the public cloud and want to extend your applications to on-premise data centers.

Pros

• Get the full benefit of managed Kubernetes on the cloud

  In this model, you get the full benefit of running the cloud provider’s managed Kubernetes on the cloud provider’s infrastructure where Kubernetes will smoothly interwork with the rest of the services.

Cons

• Vendor lock-in

  Since you are using the cloud provider’s Kubernetes both on the cloud and the on-premise data center, you have more roadblocks if you want to switch to another cloud provider.

Challenges of hybrid Kubernetes

The differences in Kubernetes distro and the cloud platforms pose some challenges specific to hybrid Kubernetes.

• Application delivery architecture

  Implementing a unified application delivery architecture is a challenge in hybrid Kubernetes. You may have to use different application delivery mechanisms for on-premise data centers and the cloud.

• IaC tools

  In hybrid Kubernetes, you will not be able to use the same IaC tools and workflows across on-premise data centers and the cloud.

• Cluster management tools

  When running different Kubernetes distros in hybrid Kubernetes, you will not be able to use the same toolset to manage all your Kubernetes clusters.

There’s no one-size-fits-all Kubernetes deployment model. You must choose what works best for your use case. But, if you can use the same Kubernetes distro on similar platforms, both in on-premise data centers and the cloud, it will ease most of your challenges in hybrid Kubernetes.

Fuga Cloud enables you to run the EMK (Enterprise Managed Kubernetes) both on the cloud and on-premise via the Fuga Cloud software solution. You can manage all your EMK clusters via the same set of tools. - Martijn Wokke, Fuga Cloud

Wrapping up

Hybrid Kubernetes can help you meet the technical and business needs that you cannot do with Kubernetes on the public cloud alone. But, hybrid Kubernetes demands more planning and more operational effort.

Choose your Kubernetes distro, the cloud platform, and the deployment model wisely, balancing the pros and cons against your requirements, to build your most successful hybrid Kubernetes architecture.

What is your experience with hybrid Kubernetes? Feel free to share your thoughts in the comments below.