The xLabs team has been collaborating with the Tanzu Edge team to build a “Bring Your Own Host” (BYOH) provider for Kubernetes. We are super excited to announce that it is now open sourced and available for tire kicking and feedback! The goal of this project is to provide a way to manage the lifecycle of Kubernetes on top of an operating system managed by the user, outside of the context of Cluster API (CAPI). This can be thought of as a bare-metal capability, or as a way to “bring your own host” or “bring your own operating system.”
The team is working with Kubernetes’ upstream SIG Cluster Lifecycle to get approval to migrate this project to the Kubernetes-SIGs GitHub repo to align with the rest of the provider community. CAPI is a Kubernetes project that automates the lifecycle of Kubernetes clusters, so you can declaratively manage your Kubernetes clusters at scale across your infrastructure. The CNCF celebrated the project reaching 1.0 in October.
By adding BYOH to the CAPI family, we are ensuring that bare-metal and edge deployments of Kubernetes can be managed in the same consistent and repeatable way you manage across your multi-cloud and hybrid infrastructure, using the existing CAPI providers for vSphere, Azure, Google Cloud, AWS, Equinix, and others. BYOH empowers you to bring all of your Kubernetes hosts under unified, multi-cloud management.
The important stuff:
- GitHub: https://github.com/vmware-tanzu/cluster-api-provider-bringyourownhost
- TGIK Demo: https://www.youtube.com/watch?v=Xwm5Ka27-Io
Here’s a good visualization of how layers of the stack are managed in a traditional CAPI context versus a BYOH CAPI context in a Tanzu Kubernetes Grid scenario.
We are already seeing interest in this solution from Kubernetes community members who were trying to solve a similar set of problems for themselves or their users. There may be patterns we can discover together as we work to get this project to a production-ready state. We are collaborating in the Kubernetes’ Slack, in the #cluster-api-baremetal and @cluster-api channels, on GitHub, and on Twitter. We invite you to share feedback and collaborate with us.
We are also excited to celebrate our collaboration with the VMware Sustainability team in building and validating this project as a carbon-reduction agent we all need in our enterprise toolboxes. By empowering the Tanzu and Kubernetes’ communities to leverage existing hardware with “Bring Your Own Host,” we’re also reducing carbon footprints. Considering that each server produces an estimated 2 kg of carbon per server annually, fewer servers is better for the planet!
|Avoided Hardware||1 Server|
|Estimated power draw per server||450 watt|
|Avoided energy per server per year (8,760 hour/year)||3,942 kilowatt-hour (kWh)|
|Total avoided energy per server per year x 1.3 (to include datacenter heat management)||5,125 kWh|
|Conversion to megawatt-hour (mWh)||5.125 mWh|
|Total annual avoided carbon using an average electricity carbon-emission factor of 0.42 kg CO2/mWh||2 kg CO2e|
A big shout-out to our Sustainability team for its ongoing work in helping us find innovative ways to fight climate change!