VMware ESXi and vSphere Cluster Management

Configure Network Failover Policies on VMware ESXi Standard Virtual Switches

Learn how to configure Notify switches, Failback, and Failover order for redundant uplinks on a VMware ESXi standard virtual switch.

Network failover policies determine how a VMware ESXi standard virtual switch responds when a physical network adapter loses connectivity. These policies apply to the physical uplinks assigned to the vSwitch and are a core part of NIC teaming and network redundancy.

This lesson covers three primary policy areas: Notify switches, Failback, and Failover order. It also explains how network failover detection fits into the process.

How ESXi Network Failover Works

An ESXi host is the physical server running the VMware hypervisor. A standard virtual switch, also called a vSwitch, is a host-local virtual switch configured directly on that host.

A physical network adapter assigned to a virtual switch is called an uplink. The uplink connects virtual machines and other virtual networking components to the physical network. A virtual machine connects through a virtual NIC, while the virtual switch selects an eligible physical uplink for external traffic.

When an uplink fails, ESXi must first determine that connectivity has been lost. This process is called network failover detection. After the failure is detected, the failover policy determines which uplink should carry traffic, whether physical switches should be notified, and what happens when the failed uplink recovers.

Detection and response are separate concepts:

Network failover detection identifies a loss of uplink connectivity.
Failover policy defines the action taken after that loss is detected.

VMware vSphere Failover Policy Settings

Policy | What It Controls | Available or Default Behavior | Operational Effect

Notify switches | Whether ESXi sends notifications after a virtual NIC connection event or uplink failover | Enabled by default | Helps physical switches refresh MAC address and forwarding information.

Failback | Whether a recovered preferred uplink immediately resumes service | Can be enabled or disabled | Controls whether traffic returns to the preferred uplink or remains on the currently working uplink.

Failover order | The priority and eligibility of physical uplinks | Each adapter is assigned as Active, Standby, or Unused | Determines which adapter carries normal traffic and which adapter is selected during failure.

Notify Switches

Notify switches is a failover setting that sends notifications when a virtual NIC becomes connected to a virtual switch and when a failover event occurs.

The purpose is to help physical switches refresh their MAC address or forwarding lookup information. If traffic begins using a different physical uplink, the physical network may need to learn that the relevant MAC address is now reachable through a new switch port.

Notify switches is enabled by default. In most redundant uplink designs, leaving it enabled helps the physical network converge more promptly after a connectivity change.

If notifications are disabled, physical switch forwarding information may not update as quickly after an uplink change. The result can be slower convergence or temporary traffic interruption, depending on the physical switch behavior and network design.

When to Check Notify Switches

After a virtual NIC connection event causes unexpected connectivity delay.
After an uplink failover takes place but traffic does not immediately reach the new path.
When investigating stale MAC address or forwarding-table information on physical switches.

Enabling this setting does not repair a failed cable, NIC, switch port, or physical network configuration. It only provides notifications associated with virtual networking changes.

Failback

Failback controls what happens after a previously failed physical uplink recovers.

Suppose one uplink is the preferred active path and a second uplink is configured as standby. If the active uplink fails, the standby uplink takes over. When the original uplink recovers, the Failback setting determines whether traffic immediately returns to it.

Failback Setting | State After Preferred Uplink Recovers | Appropriate Operational Goal

Yes | The recovered adapter immediately resumes its active role, and the replacement standby adapter returns to standby | Restore the preferred physical path as soon as it is available.

No | The recovered adapter remains out of active use until another active adapter fails | Avoid another traffic transition immediately after recovery and keep traffic on the currently functioning path.

Failback Enabled

With failback enabled, a recovered preferred adapter immediately resumes active service. The adapter that had been carrying traffic as the replacement path returns to its standby role.

This behavior is useful when the primary uplink, switch path, or network circuit is deliberately preferred and should be restored automatically.

Failback Disabled

With failback disabled, the recovered adapter does not immediately take traffic back. It remains available for a future failover event while the currently active adapter continues carrying traffic.

This behavior can reduce a second traffic transition immediately after recovery. It may be preferable when the administrator values stability on the surviving path more than automatic restoration of the original path.

Failover Order

Failover order defines the priority and eligibility of the physical adapters assigned to the standard virtual switch. Each adapter is placed into one of three roles: Active, Standby, or Unused.

Adapter Role | When It Is Used | Behavior During an Uplink Failure

Active adapters | Used while they have network connectivity during normal operation | Carry traffic and are the first uplinks considered for normal service.

Standby adapters | Reserved for use when an active adapter loses connectivity | Take over when an eligible active adapter fails.

Unused adapters | Not selected for traffic by this policy | Do not carry normal traffic and are not selected as failover paths.

Active Adapters

An Active adapter is an uplink eligible to carry traffic during normal operation. If multiple uplinks are assigned to the Active category, their exact traffic distribution also depends on the virtual switch load-balancing configuration.

Standby Adapters

A Standby adapter is reserved for a failure condition. In a simple primary-and-backup design, one uplink is Active and a second uplink is Standby. The active uplink carries traffic normally; if ESXi detects that it has lost connectivity, the standby uplink takes over.

Unused Adapters

An Unused adapter is excluded from traffic selection by the configured policy. Assigning an adapter to Unused prevents it from carrying normal traffic and from serving as a failover path for that virtual switch.

Adapter ordering should reflect the intended redundant network design. Confirm that the selected uplinks connect to appropriate physical switch ports, network segments, and redundant hardware paths.

Practical Failover Designs

Primary and Backup Uplink

Assign one physical adapter to Active adapters and a second adapter to Standby adapters.

The active uplink carries traffic normally.
Network failover detection identifies a loss of connectivity.
The standby uplink takes over.
Failback determines whether the recovered original uplink immediately resumes service.

Immediate Restoration of the Preferred Uplink

Place the preferred adapter in Active and the backup adapter in Standby, then enable Failback. After the preferred adapter recovers, it returns to active service and the backup returns to standby.

Keep Traffic on the Surviving Uplink

Use the same Active and Standby arrangement but disable Failback. After the standby adapter takes over, the recovered former active adapter does not immediately take traffic back. It remains available for a future failure event.

Exclude an Adapter from Traffic

Assign an available physical adapter to Unused adapters. The standard virtual switch will exclude it from normal traffic and failover selection.

Configure Failover Policies on an ESXi Standard Virtual Switch

Use the vSphere Web Client host networking management area to edit the standard virtual switch.

Open the vSphere Web Client.
Select the relevant ESXi host.
Open the host's Manage area.
Select the Networking tab.
Select the standard virtual switch you want to modify.
Open the virtual switch's Edit settings control.
Open the Teaming and failover section.
Set Notify switches according to the physical network design. It is enabled by default and is normally left enabled.
Set Failback to determine whether a recovered preferred uplink immediately resumes active service.
Configure Failover order by placing uplinks in Active adapters, Standby adapters, or Unused adapters.
Review the adapter roles and ordering carefully.
Save the changes.

The navigation path is:

ESXi host > Manage > Networking > select the standard virtual switch > Edit settings > Teaming and failover

Validating the Configuration

Review the resulting configuration against the intended design:

At least one intended production uplink is assigned to Active adapters.
The redundant uplink is assigned to Standby adapters if a primary-and-backup design is required.
Adapters assigned to Unused are intentionally excluded.
Notify switches matches the physical switching and convergence requirements.
Failback matches the desired recovery behavior.
Network failover detection can identify the type of uplink failure being tested.

Test failover during a suitable maintenance window. A test should confirm both the outage transition and the recovery behavior, including whether traffic returns to the original uplink or remains on the replacement uplink.

Troubleshooting Network Failover

Traffic Does Not Move to the Expected Backup Adapter

Confirm that the intended backup uplink is assigned to Standby adapters, not Unused adapters.
Verify that the expected primary uplink is assigned to Active adapters.
Review whether network failover detection identifies the loss of connectivity.
Confirm that the standby adapter is assigned to the correct standard virtual switch and has a usable physical connection.

Traffic Returns Immediately and Causes a Second Transition

Check whether Failback is enabled.
Disable Failback if the desired behavior is to keep traffic on the currently functioning adapter after recovery.

Connectivity Converges Slowly After a Failover

Verify that Notify switches is enabled.
Confirm that the physical network can process the relevant switch notifications.
Check physical switch forwarding-table behavior and uplink connectivity.
Confirm that the failure is being detected rather than only assumed from a physical symptom.

An Available Adapter Never Carries Traffic

Check whether the adapter is assigned to Unused adapters.
Review its position and role in the configured failover order.
Confirm that the adapter is assigned to the correct standard virtual switch.

Exam-Relevant Notes

Notify switches sends notifications to help physical switches update forwarding information; it does not detect the failure by itself.
Failback enabled means a recovered preferred uplink immediately resumes active service.
Failback disabled means the recovered uplink does not immediately take traffic back.
Active uplinks carry normal traffic, Standby uplinks are reserved for failover, and Unused uplinks are excluded.
Failover policy actions occur only after ESXi's network failover detection identifies an uplink connectivity failure.