Storage Networking (SAN, NAS, IP/FC) and Host Connectivity
This domain tests the path between host and LUN or share: protocol choice, zoning and authentication, multipathing, and mapping views.
SAN: FC, iSCSI, NVMe
Fibre Channel is lossless by design. It uses a switched fabric, a WWNN per node, a WWPN per port, and single-initiator, single-target zones to keep fault domains clean. iSCSI carries SCSI over TCP/IP; it needs no zoning but depends on IQN naming, CHAP authentication, VLAN separation and identical jumbo-frame settings end to end. NVMe over Fabrics removes the SCSI layer - FC-NVMe on the same fabric or NVMe/TCP on Ethernet - lowering latency and allowing far deeper queues.
NAS and the host side
File services use NFS v3/v4.1 or SMB 2.1/3.x, over DNS, LDAP/AD integration, link aggregation and VLANs. On the host, Huawei UltraPath is the recommended multipathing driver; do not let native MPIO or DM-Multipath claim the same LUNs at the same time. Access control is a chain: create the initiator, add it to a host or host group, map a LUN group through a mapping view, and have a live zone or route underneath.
Worked example
A new Linux host sees no LUN. Work outward: is the HBA port logged in to the array target port and is the zone active? Is the WWPN registered in the mapping view and in the correct host group? Is the LUN in the LUN group mapped through a port group containing that target port? Then rescan the SCSI bus and confirm multiple paths with multipath -ll, each active/optimized. The fault is normally one broken link in that chain, not the array.
Traps
Zoning several initiators into one zone; enabling CHAP on one side only; mismatched jumbo frames; leaving the host out of the host group; running UltraPath on top of native MPIO and creating duplicate paths; confusing WWNN with WWPN when registering initiators; assuming iSCSI requires FC zoning.