GitOps, Fail, Repeat – Lessons Learned From Running a Heterogeneous Platform
04-16, 12:20–12:50 (Europe/Amsterdam), The Suite

We've been using GitOps for the last three years to operate our internal managed K8s service, coming with logging, monitoring, policy enforcement, telco specific networking components and more.
Join us to learn about how we started, scaled, failed and repeated, as our environments got more heterogeneous, numerous and complex.

We will show you our first approach using just flux and directories, go into details how we outgrew it and what our pain points were.

We then settled on another design, using flux and kustomize overlays, only to outgrow that due to the heterogeneity of our environment.
We will then show you what we settled upon: versioned artifacts that contain all configuration and version details, why we think its the right fit for us but also some remaining problems. We will also show you how you can replicate our approach in your environment, if you think it might also work for you.

Max is an SRE focused on automated compute / baremetal management, CAPI Integration, PaaS components and CD for platform at Deutsche Telekom's "Das SCHIFF".

He joined DT in mid 2019 as one of the first members of the platform team and was instrumental in defining its direction, including GitOps based management. In his free time he likes to tinker with networking and embedded hardware.