VMware ESXi and vSphere Cluster Management

Fibre Channel Overview for VMware vSphere and ESXi

Learn how Fibre Channel connects ESXi hosts to shared block storage, carries SCSI traffic, supports VMFS datastores, and enables vSphere capabilities.

What Is Fibre Channel?

Fibre Channel (FC) is a dedicated, high-performance storage networking technology used primarily in enterprise environments. It transports block-storage traffic between servers and storage systems.

An FC environment is commonly called a Storage Area Network (SAN). A SAN provides servers with access to block-storage resources that appear to the operating system as disks or logical storage devices.

FC is different from general-purpose Ethernet networking at a conceptual level. Ethernet networks commonly carry many types of traffic, such as application, management, and internet traffic. FC is designed specifically for storage communication and uses specialized host adapters, switches, cabling, and storage interfaces.

SCSI Commands over Fibre Channel

Storage operations are commonly expressed using SCSI, a storage command set and model. Commands include operations such as reading blocks, writing blocks, checking device status, and discovering device information.

FC carries these storage operations between two roles:

• Initiator: The endpoint that starts storage I/O requests. In a vSphere environment, an ESXi host's FC HBA is a common initiator.
• Target: The endpoint that receives and responds to storage requests. A storage-array controller port is a common target.

FC transports encapsulated SCSI commands through the FC network. The application or virtual machine does not need to understand the details of the FC fabric; ESXi and the storage hardware handle the transport.

Core Fibre Channel Components

ESXi Host and FC HBA

A Host Bus Adapter (HBA) is a server adapter that connects an ESXi host to an FC SAN. The HBA provides the host-side FC ports and handles communication with the fabric and storage targets.

Each FC port has a unique identifier called a World Wide Port Name (WWPN). Administrators use WWPNs when configuring switch zoning and storage-array access.

FC Switch and Fabric

An FC switch is a switch designed for Fibre Channel connectivity. It connects initiator ports, such as ESXi HBA ports, to target ports on storage arrays.

An FC fabric is the interconnected FC switching infrastructure that provides paths between initiators and targets. A fabric can contain one or more switches, depending on the design.

Storage Array, Controller Ports, and LUNs

A storage array provides physical disks, flash media, controllers, and logical storage resources. Its controller ports act as storage targets.

A Logical Unit Number (LUN) is a logical block-storage device presented by the array to selected hosts. ESXi discovers a presented LUN as a storage device and can use it to create or mount a datastore.

Ports, Transceivers, and Cables

Physical FC connectivity includes HBA ports, switch ports, storage-array ports, transceivers or optics, and cables. Optical fiber is common, although supported FC implementations can also use copper cabling.

Choose the transceiver and cable type according to the installed HBA, switch, and storage hardware, as well as the required link distance. A compatible connector alone does not guarantee that a link will operate at the desired speed.

Multiple HBAs

Hosts commonly have at least two FC HBAs. Multiple adapters allow storage traffic to use separate physical paths and can protect access if one adapter, cable, switch, or fabric fails.

Fibre Channel Topologies

Point-to-Point: FC-P2P

FC-P2P connects two FC devices directly, such as one server port and one storage port. It is simple and can suit a narrow two-device use case, but it does not provide fabric-scale connectivity and has limited modern use.

Arbitrated Loop: FC-AL

FC-AL, or Fibre Channel Arbitrated Loop, connects devices in a shared loop or ring. Devices must arbitrate for access to the loop. The design has limited scalability and fault isolation, so it is largely a legacy topology.

Switched Fabric: FC-SW

FC-SW uses FC switches to connect multiple hosts and storage ports. It is the predominant enterprise design because it supports larger environments, independent paths, zoning, and better fault isolation.

Fibre Channel Media and Link Speeds

FC commonly uses optical fiber connections. Supported implementations may also use copper cabling. The selected cable and transceiver must match the capabilities of the HBA, switch port, and storage port.

16 Gbps is an example of an FC HBA link capability found in existing and legacy environments. The negotiated link speed depends on the capabilities of all relevant components, including:

• The ESXi HBA
• The FC switch port
• The optical transceiver or copper cable
• The storage-array port

A link speed is not the same as realized application throughput. Actual performance also depends on protocol overhead, path utilization, queueing, array controllers, disks or flash media, workload patterns, and virtual-machine I/O behavior.

How FC Connects ESXi to Shared Storage

The basic path from an ESXi host to an FC storage device is:

1. An application inside a virtual machine requests a storage operation.
2. The virtual machine's virtual disk maps that operation to the VMFS datastore.
3. VMFS and the ESXi storage stack issue block I/O operations.
4. The ESXi HBA sends the storage traffic through an FC fabric.
5. The storage array's target port receives the SCSI request.
6. The array processes the request against the presented LUN and returns the result through an available path.

ESXi discovers storage devices presented through FC after the adapter is connected, the fabric permits communication, and the array authorizes the host. ESXi can then associate the discovered device with one or more physical FC paths.

The relationship is important:

• A physical FC path is one route from an HBA, through fabric components, to a storage target port.
• A storage device is the logical block device, commonly a LUN, reached through one or more paths.
• A datastore is a vSphere storage container created on or mounted from that block device.

Shared FC Storage and VMFS Datastores

VMFS, the VMware Virtual Machine File System, is a clustered filesystem designed for shared block-storage datastores. An FC LUN can be formatted as a VMFS datastore, provided it is available to the intended ESXi hosts.

A VMFS datastore can contain:

• Virtual machine configuration files
• Virtual machine virtual disks
• Snapshot-related files
• Templates
• Installation media images such as ISO files

Multiple ESXi hosts can mount the same VMFS datastore. VMFS coordinates concurrent access so that cluster hosts can work with the same collection of virtual machine files.

FC and VMFS operate at different layers. FC transports block I/O through the SAN. The LUN is the block device provided by the array. VMFS organizes that block device into a datastore used by vSphere.

vSphere Capabilities Enabled by Shared Storage

vMotion

vMotion moves a running virtual machine between compatible ESXi hosts. In the common shared-storage model, both source and destination hosts must see the relevant VMFS datastore and virtual machine files. The VM's execution state moves between hosts while its files remain accessible through shared storage.

vSphere HA

vSphere High Availability (HA) restarts workloads on surviving hosts after a host failure. Shared VMFS storage gives those hosts access to the virtual machine files needed to register and power on the workloads.

DRS

Distributed Resource Scheduler (DRS) balances workloads across hosts by making placement recommendations or, according to its configuration, moving workloads. Consistent access to shared datastores is important because a destination host must be able to access the virtual machine's files.

Shared storage alone does not guarantee that these capabilities will work. Host compatibility, cluster configuration, permissions, networking, licensing, and workload requirements must also be satisfied. However, consistent datastore visibility across participating hosts is a fundamental requirement for shared-storage operations.

Availability and Multipathing

Multipathing means using multiple paths between an ESXi host and a storage device. ESXi can select an available path and, depending on the policy and hardware, distribute I/O across paths.

A resilient FC design commonly separates redundant components:

• Two or more FC HBAs in each ESXi host
• Separate cables and transceivers
• Separate FC switches or independent fabrics
• Multiple storage-array controller ports
• Multiple fabric-to-array connections

For example, one HBA can connect to Fabric A and another HBA can connect to Fabric B. The storage array can present the same LUN through target ports connected to both fabrics. ESXi then detects multiple paths to one storage device.

Separating paths across independent fabrics is more resilient than connecting both paths to the same switch. If one fabric, switch, power domain, or maintenance boundary fails, the other fabric can continue to provide access.

Typical Redundant ESXi FC Design

ESXi HBA 1 ─── Fabric A ─── Storage target port A ┐
                                                  ├── Same presented LUN
ESXi HBA 2 ─── Fabric B ─── Storage target port B ┘

In this example, both fabrics provide access to the same logical storage device. If one path fails, multipathing can use the remaining path, subject to the storage array, fabric, and ESXi configuration.

Conceptual FC SAN Configuration Workflow

1. Install and enable compatible FC HBAs in each ESXi host.
2. Connect redundant HBA ports to separate FC fabrics using compatible optics or cables.
3. Connect storage-array target ports to the appropriate fabric switches.
4. Record the WWPNs for host HBA ports and storage target ports.
5. Create FC switch zoning so the intended host initiator WWPNs can communicate with the intended storage target WWPNs.
6. Configure storage-array host registration, host groups, or LUN masking so the intended ESXi hosts can access the LUN.
7. Rescan the ESXi storage adapters after presentation or zoning changes.
8. Create a VMFS datastore on the visible LUN, or mount the existing datastore where appropriate.
9. Verify that every participating host sees the expected datastore and redundant paths.

Troubleshooting FC Storage in ESXi

ESXi Cannot See an Expected FC LUN

• Confirm the HBA is installed, enabled, and reporting an active link.
• Check the HBA, transceiver, cable, switch port, and storage port.
• Confirm FC zoning permits communication between the host WWPN and storage target WWPN.
• Confirm storage-array host registration and LUN masking include the ESXi host.
• Rescan the ESXi storage adapters after making changes.
• Verify that the intended LUN is presented and has not been claimed or configured unexpectedly.

A Shared Datastore Is Visible to Only One Host

• Compare zoning for every host in the cluster.
• Compare storage-array host access and LUN masking for every host.
• Verify that each host has access to the same LUN identifier and storage device.
• Confirm all hosts recognize the datastore consistently.
• Review each host's HBA and storage-path visibility.

Only One Path Is Available

• Inspect the second HBA, cable, optic, switch port, and storage port.
• Confirm both fabrics have the correct zoning.
• Confirm the array presents the LUN through both intended target paths.
• Review ESXi multipath status after connectivity is restored.

Storage Performance Is Below Expectations

• Verify negotiated FC speeds across the HBA, switch ports, and storage ports.
• Check for failed, inactive, or unexpectedly unused paths.
• Determine whether the workload is concentrated on a limited set of paths or array resources.
• Review array performance, queueing, and virtual-machine I/O behavior.
• Do not treat the nominal FC link speed as guaranteed application throughput.

Exam-Relevant Summary

• FC is a specialized storage networking technology for block storage.
• SCSI supplies the storage commands; FC transports those commands.
• ESXi HBA ports are common initiators, and storage-array controller ports are common targets.
• WWPNs identify FC ports and are commonly used for zoning and storage access configuration.
• A LUN is a logical block device; VMFS is the VMware clustered filesystem placed on shared block storage.
• FC-P2P is direct, FC-AL is a legacy loop design, and FC-SW is the predominant enterprise topology.
• Optics, cables, HBA ports, switch ports, and storage ports all affect link compatibility and negotiated speed.
• Multiple HBAs and independent fabrics reduce the impact of component failures.
• Multipathing provides multiple routes to a storage device and can preserve access after a path failure.
• Shared and consistently visible VMFS storage supports common vSphere operations such as vMotion, HA recovery, and DRS placement.

For related storage administration, see the Fibre Channel overview.