In this talk, I will cover how Odin, the platform for running stateful workloads at Uber, uses our custom-made distributed graph database Grail, to do declarative infrastructure at planet scale. Getting insights into our global infrastructure and supporting multiple stakeholders running databases is a challenge when you are constrained by the throughput of Git repositories and many sources contributing part of the global truth. This talk will investigate how we get near real-time answers to questions like "How many hosts of this server type do we have in this zone?" and "What is our disk utilization across the fleet?".It will also cover how we can use this capability to run a scalable, robust, and efficient storage platform through high degrees of automation, decoupling, and insights. Hopefully, you will leave this talk with some inspiration as to how you can approach getting better insights and more automation into your infrastructure.