Fibre Channel (FC) Overview for VMware ESXi

Learn how Fibre Channel connects VMware ESXi hosts to shared block storage, including SCSI transport, HBAs, fabrics, topologies, multipathing, VMFS, and vSphere capabilities.

Fibre Channel (FC) is a dedicated storage networking technology used to connect servers with block-storage systems. In a VMware vSphere environment, FC provides ESXi hosts with access to logical storage units on a storage array. ESXi can then use those devices to create shared VMFS datastores for virtual machine workloads.

FC is commonly deployed as a Storage Area Network (SAN). Unlike a general-purpose Ethernet network, an FC SAN is designed specifically for storage traffic. It provides a separate path between server storage adapters and storage-array ports.

How Fibre Channel Carries Storage I/O

Block-storage operations are represented by SCSI commands. These commands describe operations such as reading blocks, writing blocks, and discovering storage devices.

Fibre Channel transports encapsulated SCSI traffic between an initiator and a target. The initiator begins storage I/O requests; in an ESXi host, the initiator is typically an FC Host Bus Adapter (HBA). The target receives those requests and provides storage; it is typically an FC port on a storage array.

The array presents one or more logical block-storage units called LUNs. ESXi discovers eligible LUNs and can use them directly or format them as VMFS datastores.

Core Components in an ESXi FC Storage Path

Component	Location	Role	Example Relationship
ESXi host	Server	Runs virtual machines and issues storage I/O	Uses an HBA to send SCSI requests
FC HBA	ESXi host	Provides the physical and FC protocol connection to the SAN	Acts as the host-side initiator
FC cable or transceiver	Between devices	Carries optical or supported copper FC signals	Connects an HBA port to a switch port
FC switch	SAN fabric	Interconnects and forwards traffic between FC devices	Connects host HBAs to array ports
Storage-array FC port	Storage system	Provides the target-facing connection	Receives requests for presented LUNs
LUN	Storage array	Logical block-storage unit presented to hosts	Can be visible to multiple ESXi hosts
VMFS datastore	vSphere storage layer	Clustered filesystem used by ESXi hosts	Stores virtual machine and related files

Host Bus Adapters

An FC HBA is an adapter installed in an ESXi server. It provides the port, electronics, and protocol processing needed to communicate with the FC SAN. The HBA is different from an ordinary Ethernet network adapter because it is designed for FC storage communication.

Hardware, firmware, drivers, optics, and switch compatibility must be checked as a complete combination. An HBA may advertise a particular link capability, but the negotiated connection can be lower because of the switch, optic, cable, or storage port.

FC Switches and Fabrics

An FC switch creates and forwards connections within an FC fabric. A fabric is the interconnected FC switching environment that joins initiators and targets. In a larger deployment, hosts and storage systems are connected through one or more fabrics rather than directly to each other.

FC switch zoning controls which initiators can communicate with which targets. Storage-array access controls, often called LUN masking or host access configuration, then determine which LUNs are presented to each ESXi host.

Cabling and Transceivers

FC deployments often use optical transceivers and fiber cabling, especially for connections between racks or across longer data-center distances. Supported copper connections can also be appropriate for particular short-distance deployments.

Validate the compatibility of the cable, transceiver, HBA, switch port, and storage-array port together. Matching the nominal speed alone does not guarantee a supported or operational link.

Redundancy, Independent Paths, and Multipathing

ESXi hosts commonly use two or more FC HBAs so that a failure of one adapter, cable, switch, or storage connection does not necessarily interrupt storage access.

A resilient design uses independent paths. For example, one HBA can connect to one FC fabric while a second HBA connects to a separate fabric. The storage array exposes the same LUN through appropriate controller or target ports in both fabrics.

ESXi detects these routes as multiple paths to the same storage device. Multipathing manages those paths for availability and, depending on the selected policy and array behavior, path use. The paths are normally not treated as separate datastores; they are alternate connections to one device and one datastore.

ESXi host A                    FC SAN                         Storage array
  HBA 1  ----------------  Fabric A  ----------------  Controller port A
  HBA 2  ----------------  Fabric B  ----------------  Controller port B
                                                        Same presented LUN

Several ESXi hosts can discover that LUN and mount it as one shared VMFS datastore.

Typical Dual-Fabric Example

An ESXi host has two FC HBAs.
Each HBA connects to a separate FC switch fabric.
The array exposes the same LUN through target ports in both fabrics.
ESXi discovers multiple paths to one storage device.
Multipathing maintains access if an individual path or fabric fails.

Redundancy improves storage availability, but it must be designed end to end. Two HBAs connected to the same failed switch do not provide the same protection as two HBAs connected to independent fabrics.

FC Link Speed and Effective Throughput

Fibre Channel is designed for high-throughput, low-latency storage traffic. A 16 Gbps HBA link is an example of a supported link capability in some environments; it is not a universal current limit. FC generations and device capabilities vary.

An advertised link speed is the signaling rate of the connection. Effective workload throughput is usually lower and depends on protocol overhead, I/O size, queue depth, the number of active paths, fabric congestion, storage-array performance, and workload contention. A fast HBA cannot make an overloaded array deliver data faster than the array can process it.

Fibre Channel Topologies

Topology	Connection Model	Typical Scale	Modern Usage	Key Consideration
Point-to-point (FC-P2P)	Direct connection between two FC devices	Very limited	Specialized or simple two-device arrangements	Does not provide fabric-level expansion or shared switching
Arbitrated loop (FC-AL)	Devices share a ring or loop and arbitrate for access	Limited	Mostly older or specialized designs	A shared loop can limit scalability and fault isolation
Switched fabric (FC-SW)	Hosts and storage connect through FC switches	Scalable from small to enterprise deployments	Typical topology for modern storage arrays	Requires switch design, zoning, and path planning

Point-to-Point

FC-P2P connects two FC devices directly, such as one host-side port and one storage-side port. It can fit a limited arrangement, but adding more hosts, storage ports, or alternate paths quickly becomes impractical without switching.

Arbitrated Loop

FC-AL connects participating devices in a shared loop. Devices arbitrate for use of the loop rather than communicating through a switched fabric. This topology has limited modern use because shared access and loop behavior make expansion and fault isolation less convenient.

Switched Fabric

FC-SW uses FC switches to create a fabric. It supports multiple hosts, storage controllers, and independent paths, making it the normal topology for enterprise storage-array deployments.

How FC Storage Becomes a vSphere Datastore

A storage array presents a LUN to one or more ESXi hosts. After the hosts discover the LUN, an eligible device can be formatted with VMFS, VMware's clustered filesystem for shared ESXi datastores.

A datastore is a vSphere storage container. A VMFS datastore can contain virtual machine configuration files, virtual disks, snapshots where applicable, virtual machine templates, and installation media such as ISO images.

Because VMFS is designed for coordinated access by multiple ESXi hosts, several hosts can mount the same datastore. The hosts must have consistent and correctly controlled access to the same LUN, and the storage presentation must be configured safely.

Shared FC Storage and vSphere Features

vSphere Feature	How Shared FC Storage Helps	Additional Requirements to Acknowledge
vMotion	A running virtual machine can remain on a datastore accessible to both the source and destination hosts while its compute execution moves between compatible hosts.	Compatible hosts, vMotion networking, CPU and configuration compatibility, licensing, and correct cluster configuration are also required.
VMware HA	If a host fails, a virtual machine stored on shared storage can be restarted on a surviving host that can access its files.	HA configuration, admission control, host health, networking, capacity, licensing, and other cluster requirements also apply.
DRS	Shared datastores give DRS-managed hosts a common location from which workloads can be placed or balanced.	DRS licensing, cluster configuration, compute capacity, compatibility, networking, and policy settings also affect operations.

Planning and Configuration Concepts

Install supported FC HBAs in each ESXi host.
Connect the HBAs using compatible optics and cabling.
Build independent FC paths, preferably through separate fabrics and appropriate array target ports.
Configure FC switch zoning so the intended initiators can reach the intended targets.
Register the hosts and configure LUN masking or host access on the storage array.
Rescan the ESXi storage adapters after the presentation is configured.
Confirm that the expected LUN and all redundant paths are visible.
Create or mount a VMFS datastore from the accessible FC LUN.
Review the ESXi multipathing policy and verify path states.

In a cluster, apply storage presentation consistently to every intended host. A datastore that is visible to only some hosts cannot provide the same shared-storage behavior for the entire cluster.

Troubleshooting FC Storage in ESXi

ESXi Does Not See a Newly Presented LUN

Check HBA link status and driver or firmware compatibility.
Verify physical cabling and transceiver compatibility.
Review FC switch zoning.
Check storage-array host registration and LUN masking.
Confirm that the ESXi host completed a storage rescan.

Only One Path Appears

Check the second HBA and its connection to the second fabric.
Verify independent zoning for each fabric.
Confirm that the target ports are enabled on both paths.
Ensure both paths present the same LUN identity rather than different devices.
Review ESXi path discovery and multipathing configuration.

A Shared Datastore Is Visible to Only One Host

Verify consistent LUN presentation to every intended ESXi host.
Compare host-specific zoning and array access permissions.
Check the VMFS datastore mounting state on each host.
Confirm that the hosts are connected to the same planned storage fabrics.

Storage Performance Is Lower Than Expected

Check negotiated link speed instead of relying on nominal HBA capability.
Look for path imbalance or failed paths.
Check for oversubscription or congestion in the FC fabric.
Measure storage-array latency and workload contention.
Verify HBA, switch, and array firmware interoperability.

Exam-Relevant Summary

FC is a dedicated storage networking technology for block-storage traffic, not a general-purpose Ethernet network.
ESXi HBA ports are typically initiators; storage-array FC ports are typically targets.
FC transports SCSI commands between initiators and targets.
A LUN is a logical block-storage unit presented by an array.
VMFS is a clustered filesystem that allows multiple ESXi hosts to access a shared datastore.
FC-P2P is direct, FC-AL is loop-based, and FC-SW uses switches.
Switched fabric is the usual topology for modern enterprise FC deployments.
Multiple paths to one LUN are managed with multipathing and are not normally separate datastores.
Optics, cables, HBAs, switches, and storage ports must be validated as a compatible set.
Shared FC storage supports vSphere operations only when their additional licensing, networking, compatibility, capacity, and configuration requirements are satisfied.

For related administration topics, see the VMware ESXi online course, adding a raw LUN, vSphere HA admission control, and datastores in vSAN for comparison with another shared-storage architecture.