Storage Spaces Direct is the software-defined infrastructure that powers Azure Stack and Azure Stack HCI. This session is the second half (part 2 of 2) of the definitive "How It Works" guide, including the undocumented nitty-gritty details, straight from the engineering team. What types of drives can I use? How does the cache work? How does Storage Spaces Direct rely on Ethernet and RDMA networking? What's the end-to-end IO path? What's this about mirror and parity? How much space will that give me? And what on Earth is mirror-accelerated parity?! How many failures is too many failures? What about server failures? What even is a fault domain? Why do servers in my cluster need to resync? Is that different than repairing? What's rebalancing...? All this and more!