How do we merge storage networks into leaf-and-spine fabric (40/100G)? It makes sense regarding throughput, but poses additional risks as opposed to separate FC SAN: fabric outage causes disks to stop responding.