VMware ESXi and vSphere Cluster Management

Fibre Channel Components in VMware ESXi

Learn how ESXi hosts, HBAs, Fibre Channel switches, storage processors, RAID, and LUNs work together in a redundant FC SAN.

Fibre Channel (FC) is a high-speed storage networking protocol used to connect servers to shared block storage. In a VMware environment, an ESXi host uses FC connectivity to access storage devices supplied by a disk array. ESXi can then use those devices for datastores and virtual machine disks.

This lesson explains the components of a native Fibre Channel storage area network (SAN), the path from an ESXi host to a LUN, and how redundancy keeps storage available when a component fails.

Fibre Channel SAN overview

A Storage Area Network (SAN) is a dedicated storage network that provides hosts with access to shared block storage. In an FC SAN, the storage traffic uses Fibre Channel rather than ordinary Ethernet networking.

A typical VMware FC SAN contains:

• ESXi hosts: Servers that run virtual machines and consume shared storage.
• Host bus adapters (HBAs): Host-side adapters that connect ESXi servers to the FC fabric.
• FC switches: Fibre Channel-aware switches that forward storage traffic between SAN nodes.
• Storage processors (SPs): Storage-array controllers that manage access to logical storage.
• Storage systems: Disk arrays containing physical disks and controller hardware.
• Logical units: Logical storage devices, commonly identified by LUNs, that are presented to hosts.

The end-to-end path is generally:

ESXi host → HBA → FC switch or fabric → storage processor port → storage system → presented LUN

The storage array creates the LUN from its available capacity. The storage administrator then authorizes selected host connections to access it. ESXi discovers the authorized LUN as a storage device and can use it to create or extend a datastore, depending on the storage design.

Core Fibre Channel SAN components

Storage systems and physical capacity

A storage system, also called a disk array, is a collection of physical disks managed by intelligent controller hardware. The controllers provide functions such as capacity management, RAID processing, logical-unit creation, host access control, and communication with storage networks.

The physical disks are not normally presented directly to ESXi. Instead, the storage system organizes their capacity and creates logical storage devices. This separation allows the array to manage disk layout, redundancy, performance, and access policy while hosts consume a consistent block-storage interface.

JBOD and RAID organization

JBOD means “Just a Bunch Of Disks.” The term describes a group of disks without implying a particular RAID configuration. Depending on the array and its design, capacity may be used as individual disks, disk groups, pools, or other managed arrangements.

A RAID set is a group of disks organized for redundancy, performance, or both. For example, a RAID arrangement may continue operating after a disk failure, although the exact protection depends on the RAID level and implementation.

The usual conceptual order is:

1. Physical disks provide raw capacity.
2. The array organizes disks into a disk group, pool, or RAID set.
3. The storage system carves part of that capacity into one or more logical units.
4. The storage system presents selected logical units to authorized host connections.
5. ESXi discovers the presented devices and uses them for storage, such as a VMFS datastore.

RAID and disk layout are storage-array concerns. A LUN is the logical device exposed to hosts; it is not the same thing as an individual physical disk or RAID set.

Logical units and LUN presentation

LUN means Logical Unit Number. In practical FC administration, a LUN identifies a logical storage device that a host can address through the storage network. The device is created from storage capacity such as a RAID-backed disk group or disk pool.

The storage array uses LUN masking to control visibility. LUN masking restricts a logical unit so that only authorized host connections can access it. For example, an administrator can present a production LUN to the HBAs of a production ESXi cluster while withholding it from an unrelated test host.

After the array presents the LUN and the FC paths are operational, ESXi discovers the LUN as a storage device. The device can then be used in the appropriate storage workflow, such as creating or mounting a datastore. A guest operating system normally accesses a virtual disk backed by that datastore rather than addressing the LUN directly.

Storage processors

A storage processor (SP) is a controller in the storage array. It receives storage requests from host connections, manages access to logical units, and coordinates those requests with the disks or pools that provide the capacity.

The relationship between the main storage elements is:

• JBOD or physical disks provide the underlying capacity.
• RAID sets or disk pools organize that capacity.
• Storage processors manage the organized capacity.
• The storage system creates LUNs from that capacity.
• Storage processors present the LUNs through their storage ports and enforce host-access rules.

Many arrays include more than one storage processor. Multiple controllers and ports can provide alternate paths to the same logical unit. The exact ownership and failover behavior depends on the storage-array model, but the design goal is continued access when a controller, port, cable, or fabric path fails.

Host bus adapters

An HBA, or Host Bus Adapter, is the adapter installed in an ESXi server that provides Fibre Channel connectivity. In FC terminology, the HBA acts as a host-side initiator: it initiates storage communication with target ports on the storage system.

An HBA connects to the SAN through an FC link, commonly using fiber cabling and an appropriate transceiver. ESXi uses the adapter and its driver to discover FC targets and the LUNs those targets make available.

Fault-tolerant designs commonly use at least two HBAs per ESXi host. Each HBA should use an independent path to the storage environment. For example:

• HBA 1 connects through FC fabric A.
• HBA 2 connects through FC fabric B.
• Each fabric reaches an appropriate storage processor port.
• Both paths can reach the same presented LUN.

Two adapters alone do not guarantee redundancy. The cables, switch paths, storage ports, controller paths, and ESXi multipathing configuration must also be designed and configured without a shared failure point.

Fibre Channel switches and the SAN fabric

An FC switch is a Fibre Channel protocol-aware switching device. Its broad purpose resembles that of an Ethernet switch: it receives traffic on one port and forwards it toward the correct destination. However, an FC switch is specialized for storage traffic and FC identifiers, links, and fabric behavior.

The switches connect SAN nodes, including:

• FC ports on ESXi host HBAs.
• FC ports on storage processors.
• Other FC devices when included in the SAN design.

The collection of interconnected FC switches, ports, and links is the SAN fabric. The fabric provides the transport path between host initiators and storage targets. Redundant fabrics or independent switch paths allow traffic to continue when one switch, link, adapter, or storage port becomes unavailable.

Connectivity through a switch does not by itself authorize access to every LUN. Fabric connectivity and storage-side access control are separate concerns. The host must have a working path, and the storage system must present the requested LUN to that host connection.

Typical redundant FC topology

                         ┌─────────────────────────────┐
                         │       Storage system        │
                         │  RAID or disk pool capacity │
                         │   LUN 1       LUN 2          │
                         │  SP-A ports   SP-B ports    │
                         └───────┬───────────┬─────────┘
                                 │           │
                          Fabric A       Fabric B
                                 │           │
                    ┌────────────┘           └────────────┐
                    │                                     │
              ┌─────┴─────┐                         ┌─────┴─────┐
              │ FC switch A│                         │ FC switch B│
              └─────┬─────┘                         └─────┬─────┘
                    │                                     │
                 HBA 1                                 HBA 2
                    └──────────────┬──────────────────────┘
                                   │
                              ESXi host
                                   │
                         VMFS datastore or device
                                   │
                           Virtual machine disk

In this design, the storage array creates and presents the LUNs, the HBAs initiate access, and the FC switches carry traffic between the host and storage. If an adapter, cable, switch, or one storage path fails, another path can continue serving the same LUN, provided multipathing and the storage platform support the required failover behavior.

Responsibilities along the path

ESXi and virtual machine storage visibility

ESXi uses the FC SAN to access shared block devices. At the ESXi layer, the host discovers storage devices and maintains paths to them. The host can then provide shared storage to virtual machines through a datastore or another configured storage model.

A virtual machine normally sees a virtual disk and standard SCSI-style storage semantics. SCSI is a storage command model used by operating systems and virtual machines; it describes how storage requests are represented, not necessarily the physical transport carrying them. The physical path may use Fibre Channel even though the guest sees a virtual SCSI disk.

FC configuration, LUN presentation, path redundancy, and multipathing are normally managed at the ESXi host and storage-infrastructure layers. Individual guest operating systems generally do not need to configure the underlying FC switches, HBAs, or array paths.

Storage discovery and high-level configuration order

No command procedure is required to understand the component model. In a real deployment, configuration is usually coordinated between the storage and virtualization administrators.

1. Install and configure compatible FC HBAs in the ESXi host.
2. Connect the HBAs to independent FC switch or fabric paths.
3. Connect the redundant fabric paths to storage processor ports.
4. Organize storage capacity using disks, JBOD, RAID sets, or pools.
5. Create a LUN from the available storage capacity.
6. Authorize the intended ESXi host HBA connections using storage-side presentation or masking.
7. Verify that the expected FC paths are active.
8. Rescan storage from the ESXi host or vSphere management interface so newly presented devices are discovered.
9. Use the discovered device in the datastore configuration workflow.

Always follow the storage vendor's compatibility, multipathing, and presentation requirements. A LUN should be presented consistently to all hosts that are intended to share the same datastore.

Troubleshooting component relationships

ESXi cannot see an expected LUN

• Check whether the LUN was created on the storage system.
• Confirm that the LUN is presented or masked to the host's HBA connections.
• Verify that the relevant HBA is online and has the expected link state.
• Check cables, transceivers, and FC switch ports.
• Confirm that the storage processor port has an active path to the host fabric.
• Rescan storage from ESXi after the array-side presentation is complete.

The LUN is visible but has only one path

• Check whether one HBA is offline, disconnected, or incorrectly configured.
• Inspect the redundant FC switch or fabric path.
• Check the corresponding storage processor port and controller path.
• Review physical cabling for a shared failure point.
• Review ESXi multipathing and the storage array's supported path policy.

A virtual machine administrator expects to configure FC in the guest

Explain that the FC connection is normally handled by ESXi and the SAN. The guest usually consumes a virtual SCSI-style disk backed by an ESXi datastore. Direct guest access to FC hardware is a separate design and is not the normal model for virtual disks stored on an FC-backed datastore.

Fibre Channel and FCoE

Fibre Channel over Ethernet (FCoE) carries Fibre Channel traffic over an Ethernet-based network. It is related to native FC but uses different infrastructure choices.

This lesson focuses on native Fibre Channel component roles. FCoE requires its own compatibility, adapter, switching, and operational considerations.

Key takeaways

• FC is a storage networking protocol for shared block storage.
• An ESXi host connects to the FC fabric through one or more HBAs.
• FC switches provide fabric connectivity between host HBAs and storage processor ports.
• A storage system uses disks, JBOD, RAID sets, or pools to provide capacity.
• Storage processors create and present logical devices called LUNs.
• LUN masking limits which host connections can access a logical unit.
• ESXi discovers presented LUNs as storage devices and can use them for datastores.
• Virtual machines normally see virtual SCSI-style disks rather than the underlying FC components.
• Redundant HBAs, fabrics, cables, storage ports, and multipathing provide continued access after a component failure.

For a concise component reference, see Fibre Channel Components.