NVMe-over-Fabrics (NVMe-oF) with Windows Server Initiator and Everpure FlashArray

We have been using NVMEoFC for 3 years now with Linux pointed to our C arrays. We are hoping to get AC stretching implemented for NVMEoFC on our new XL170s. We need the best queue depth possible. Traditional FC has a QD of 32 and NVME has a QD of 65000+ queues, with each queue supporting 65000+. It is magnitudes deeper. Good to hear that MS is now pushing to that end. 

I believe most people have the infrastructure for it now, but don't understand the benefits.