Giant Swarm provides managed Kubernetes clusters for our customers, both on-premises and in the cloud.
On-prem, these Kubernetes clusters run entirely on a ‘host’ Kubernetes cluster, which have api servers inaccessible from the Internet. This poses interesting problems for continuous deployment. We introduced a new CI/CD process that manages continuously deploying to on-prem clusters, including a new open-source in-cluster service that manages these deployments.
This talk focuses on our process of deploying infrastructure services to these on-prem datacenters, as well as our overall release process and the challenges of continuously deploying modern infrastructure as a distributed startup.
Required audience experience
Objective of the talk
To present how we at Giant Swarm deploy Kubernetes infrastructure to on-premises datacenters.