Ever wondered how storage services are run at petabyte scale at CERN? In this talk, we will explore the journey we started last year when we began embracing Infinite Scale as the replacement for our traditional ownCloud X deployment. We will talk about some of the design choices and patterns we adopted to achieve feature-parity with our previous system, while keeping everything easy to operate, monitor and troubleshoot. We’ll reveal the reasons behind our choice for a bare-metal deployment on our production instance, while keeping alive a full stack version of our entire service ecosystem reproduced on Kubernetes for ScienceBox. And last, we will reflect on the challenges posed by running a distributed microservices architecture in the real world and some common mistakes that you can avoid when facing a similar choice.