At Giant Swarm, Platform Engineer Marcel Müller is on the team that manages Kubernetes clusters for their customers. That's a lot of Kubernetes clusters — a whole lot. From early on, they invested heavily in automation to be able to handle this large lot of clusters efficiently. Their weapons of choice are Kubernetes operators, which they develop and run in their environments. These operators enable them to manage clusters in a fault-tolerant and highly resilient way.
There is a catch though: by building their own operators, they had introduced their own declarative API for cluster lifecycle management. This API is very different from the approaches of other providers. Therefore, switching between different providers has become difficult for users.
Thankfully, the upstream [cluster API] project introduces a community-driven approach to a declarative Kubernetes-style API for cluster lifecycle management. In this talk, Marcel presents Giant Swarm’s journey so far in building up their architecture with Kubernetes operators, as well as describe their progress and future vision of adopting the upstream cluster API.