CloudEngine 9800, 8800, 6800, and 5800 Series

Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3 Link Aggregation Configuration

Link aggregation is a technology that bundles multiple Ethernet links into a logical
link to increase bandwidth, improve reliability, and load balance traffic.

3.1 Overview of Link Aggregation

3.2 Understanding Link Aggregation
3.3 Application Scenarios for Link Aggregation
3.4 Summary of Link Aggregation Configuration Tasks
3.5 Licensing Requirements and Limitations for Ethernet Link Aggregation
3.6 Default Settings for Link Aggregation
3.7 Configuring Link Aggregation in Manual Load Balancing Mode
3.8 Configuring Link Aggregation in LACP Mode
3.9 Configuring Preferential Forwarding of Local Traffic on an Eth-Trunk in a Stack
On a network where interfaces of multiple switches in a stack form an Eth-Trunk,
you can configure the Eth-Trunk to preferentially forward local traffic to increase
bandwidth use efficiency between stack devices and improve traffic forwarding
efficiency.
3.10 Creating an Eth-Trunk Layer 3 Sub-interface
3.11 Maintaining Link Aggregation
3.12 Configuration Examples for Link Aggregation
3.13 Troubleshooting Link Aggregation

3.1 Overview of Link Aggregation

Definition
Ethernet link aggregation, also called Eth-Trunk, bundles multiple physical links
into a logical link to increase link bandwidth. The bundled links back up each
other, increasing reliability.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 79

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Purpose
As networks grow in scale, users require Ethernet backbone networks to provide
higher bandwidth and reliability. In the past, to increase the bandwidth, users had
to replace legacy devices with high-speed devices. This solution, however, is costly
and inflexible.

Link aggregation helps increase bandwidth by bundling a group of physical

interfaces into a single logical interface, without having to upgrade hardware. In
addition, link aggregation provides link backup mechanisms, greatly improving
link reliability.

Link aggregation has the following advantages:

● Increased bandwidth
The bandwidth of the link aggregation interface is the sum of the bandwidth
of member interfaces.
● Higher reliability
When an active link fails, traffic on this active link moves to another active
link, improving reliability of the link aggregation interface.
● Load balancing
In a link aggregation group (LAG), traffic is load balanced among active link
members.

3.2 Understanding Link Aggregation

3.2.1 Concepts
In Figure 3-1, DeviceA and DeviceB are connected through three Ethernet physical
links. These links bundle into a logical link, and their bandwidths are combined to
form the total bandwidth of the logical link. The three physical Ethernet links
provide backup for each other, improving reliability.

NOTE
Both devices connected by the Eth-Trunk must use the same number of physical interfaces,
interface rate, jumbo, and flow control mode.

Figure 3-1 Eth-Trunk networking

The link aggregation interface can be used as a common Ethernet interface to

implement routing protocols and other services. Unlike a common Ethernet
interface, the link aggregation interface needs to select one or more member
interfaces to forward traffic.

Link aggregation concepts are described as follows:

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 80

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Link aggregation, link aggregation group (LAG), and link aggregation

interface
Link aggregation technology bundles a group of physical interfaces into a
logical interface to increase bandwidth and improve reliability.
An LAG binds multiple Ethernet links.
Each LAG has one logical interface, that is, link aggregation interface or Eth-
Trunk.
● Member interface and member link
The interfaces that constitute an Eth-Trunk are member interfaces. A link
corresponding to a member interface is a member link.
● Active and inactive interfaces and links
There are two types of interfaces in a LAG: active interfaces that forward data
and inactive interfaces that do not forward data.
The link connected to an active interface is the active link, and the link
connected to an inactive interface is the inactive link.
● Upper threshold for the number of active interfaces
When the number of active interfaces reaches this threshold, the bandwidth
of the Eth-Trunk will not increase even if more member links go Up. This
guarantees high network reliability. When the number of active member
interfaces reaches the upper threshold, additional active member interfaces
go Down.
For example, 8 fully-functioning member links bundle into an Eth-Trunk link,
with each link providing a bandwidth of 1 Gbit/s. If the Eth-Trunk link only
needs to provide a maximum bandwidth of 5 Gbit/s, you can set the
maximum number of Up member links to 5 or larger. The remaining
unselected links in Up state automatically enter the backup state, improving
reliability.
NOTE

The upper threshold for the number of active interfaces is inapplicable to the manual
load balancing mode. For details about the manual load balancing mode, see 3.2.3
Link Aggregation in Manual Load Balancing Mode.
● Lower threshold for the number of active interfaces
When the number of active interfaces falls below the lower threshold, the
Eth-Trunk goes Down. This guarantees the minimum available bandwidth for
the Eth-Trunk.
For example, if the Eth-Trunk is required to provide a minimum bandwidth of
2 Gbit/s and each member link's bandwidth is 1 Gbit/s, the lower threshold
must be set to 2 or larger.

3.2.2 Forwarding Principle

As shown in Figure 3-2, an Eth-Trunk is deployed at the data link layer, that is,
between the LLC sub-layer and the MAC sub-layer.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 81

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Figure 3-2 Eth-Trunk in the Ethernet protocol stack

The Eth-Trunk forwarding table is composed of the following entries:

● HASH-KEY value
The key value is calculated through the hash algorithm based on the MAC
address or IP address in a data packet.
● Interface number
Eth-Trunk forwarding entries are relevant to the number of member interfaces
in an Eth-Trunk. Different HASH-KEY values map different outbound
interfaces.
If physical interfaces 1, 2, 3, and 4 are bundled into an Eth-Trunk, the Eth-
Trunk forwarding table contains four entries and the HASH-KEY values
correspond to interface numbers, as shown in Figure 3-3.

Figure 3-3 Example of an Eth-Trunk forwarding table

The Eth-Trunk module forwards a packet according to the Eth-Trunk forwarding

table:
1. The Eth-Trunk module receives a packet from the MAC sub-layer, and then
extracts its source MAC address/IP address or destination MAC address/IP
address.
2. The Eth-Trunk module calculates the HASH-KEY value using the hash
algorithm.
3. Based on the HASH-KEY value, the Eth-Trunk module searches the Eth-Trunk
forwarding table for the interface number, and then sends the packet from
the corresponding interface.

3.2.3 Link Aggregation in Manual Load Balancing Mode

Link aggregation can work in manual load balancing or Link Aggregation Control
Protocol (LACP) mode. The main difference between the two modes is whether
LACP is used.
In manual load balancing mode, you must manually create an Eth-Trunk and add
member interfaces to the Eth-Trunk. LACP is not used. In this mode, all active links
load balance traffic evenly. If an active link fails, the other active links share the

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 82

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

traffic evenly. The manual load balancing mode is suitable in scenarios where the
link between two directly connected devices requires a high bandwidth but the
devices do not support the LACP protocol.

3.2.4 Link Aggregation in LACP Mode

Background
An Eth-Trunk in manual load balancing mode can increase the bandwidth. In
manual mode, however, only member link disconnections can be detected. Other
faults, such as link layer faults and incorrect link connections, cannot be detected.
The Link Aggregation Control Protocol (LACP) can be used to detect more fault
types, improving fault tolerance of the Eth-Trunk, providing backup, and ensuring
high reliability of member links.
LACP provides a standard negotiation mechanism that a switching device can use
to create and start the aggregated link based on its configuration. LACP maintains
the link status after the aggregated link is created and adjusts or removes the link
if an aggregated link's status changes.
For example, in Figure 3-4, four interfaces on DeviceA should be connected to the
corresponding interfaces on DeviceB, and these interfaces are all bundled into an
Eth-Trunk. However, one interface on DeviceA has been connected to an interface
on DeviceC. As a result, DeviceA may send data destined for DeviceB to DeviceC. If
link aggregation in manual mode load balancing is used, this fault would go
undetected.
In this situation, if LACP is enabled on DeviceA and DeviceB, the Eth-Trunk only
selects active links (links connected to DeviceB) to forward data after negotiation.
Data sent by DeviceA destined for DeviceB only reaches DeviceB.

Figure 3-4 Incorrect Eth-Trunk connection

Concepts
● LACP system priority
LACP system priorities determine the sequence in which devices at two ends
of an Eth-Trunk select active interfaces to join a LAG. In order for a LAG to be
established, both devices must select the same interfaces as active interfaces.
To achieve this, one device (with a higher priority) is responsible for selecting
the active interfaces. The other device (with a lower priority) then selects the
same interfaces as active interfaces. In priority comparisons, numerically
lower values have higher priority.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 83

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● LACP interface priority

Interface LACP priorities affect which interfaces of an Eth-Trunk are selected
as active interfaces. The smaller the LACP interface priority value, the higher
the LACP interface priority. The interfaces with the highest LACP interface
priority become active interfaces.

Implementation of Link Aggregation in LACP Mode

LACP, as specified in IEEE 802.3ad, implements dynamic link aggregation and de-
aggregation, allowing devices at both ends of the link to exchange Link
Aggregation Control Protocol Data Units (LACPDUs).
After member interfaces are added to an Eth-Trunk in LACP mode, each device
sends LACPDUs to inform the other device of its system priority, MAC address,
member interface priorities, interface numbers, and keys. Keys are used to
determine whether the remote end connected to each interface is in the same
LAG and whether bandwidth of each interface is the same. The other device then
compares this information with its own corresponding information, and selects
which interfaces are to be aggregated. Both devices perform LACP negotiation to
select active interfaces and links.
● An Eth-Trunk in LACP mode is set up as follows:
a. Devices at both ends exchange LACPDUs.
As shown in Figure 3-5, create an Eth-Trunk in LACP mode on DeviceA
and DeviceB and add member interfaces to the Eth-Trunk. Then enable
LACP on the member interfaces. Both devices can then exchange
LACPDUs.

Figure 3-5 Exchange of LACPDUs

b. Devices at both ends determine the Actor and active links.

As shown in Figure 3-6, when DeviceB receives LACPDUs from DeviceA,
DeviceB checks and records information about DeviceA and compares
system priorities. If the system priority of DeviceA is higher than that of
DeviceB, DeviceA becomes the Actor. If DeviceA and DeviceB have the
same system priority, the device with a smaller MAC address becomes the
Actor.
After devices at both ends select the Actor, they select active interfaces
according to the priorities of the Actor's interfaces. An Eth-Trunk is
established when devices at both ends select the same interfaces as
active interfaces. After Eth-Trunk is established, active links load balance
data.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 84

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Figure 3-6 Selecting the Actor and active links in LACP mode

● LACP preemption
When LACP preemption is enabled, interfaces with higher priorities in a LAG
always function as active interfaces.
As shown in Figure 3-7, Port 1, Port 2, and Port 3 are member interfaces of
an Eth-Trunk; DeviceA acts as the Actor; the upper threshold for the number
of active interfaces is 2; LACP priorities of Port 1, Port 2, and Port 3 are 10, 20,
and 30 respectively. When LACP negotiation is complete, Port 1 and Port 2 are
selected as active interfaces because their LACP priorities are higher, and Port
3 is used as the backup interface.

Figure 3-7 LACP preemption

LACP preemption has the following effects on selection of the active

interfaces:
– LACP preemption allows the original active interface to be re-selected
after recovering from a fault. For example, when Port 1 fails, Port 3
replaces Port 1 as the active interface. LACP preemption is not enabled
on the Eth-Trunk by default, and Port 1 remains in the backup state after

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 85

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

it recovers. If LACP preemption is enabled on the Eth-Trunk, Port 1

replaces Port 3 to become the active interface again.
– LACP preemption allows active interfaces to be re-selected when the
LACP interface priority changes. For example, the LACP interface priority
of Port 3 is changed to 5. If LACP preemption is enabled, Port 3 will
replace Port 2 as an active interface.
● LACP preemption delay
The LACP preemption delay is the time that a backup link waits before
becoming the active link after LACP preemption occurs. The LACP preemption
delay is used to prevent unstable data transmission over an Eth-Trunk link
caused by frequent status changes of member links.
● Switchover between active and inactive links
In LACP mode, a link switchover in a LAG is triggered if a device at one end
detects one of the following events:
– An active link goes Down.
– LACP detects a link fault.
– An active interface becomes unavailable.
– When LACP preemption is enabled, a backup interface becomes the
active interface when its priority is changed to be higher than that of the
current active interface.
When any of the preceding events occurs, LACP takes effect in the following
sequence:
a. Shuts down the faulty link.
b. Selects the backup link with the highest priority among N backup links to
replace the faulty active link.
c. The highest priority backup link becomes the active link and begins
forwarding data.

LACP Implementation Modes

LACP can work in static or dynamic LACP mode:

● Static LACP mode

In static LACP mode, two ends exchange LACPDUs to negotiate link
aggregation parameters to determine active and inactive interfaces.
In static LACP mode, you must manually create an Eth-Trunk and add
member interfaces to the Eth-Trunk. Different from the manual load
balancing mode, the static LACP mode selects active member interfaces by
sending LACPDUs. That is, when a group of interfaces are added to an Eth-
Trunk, devices at the two ends determine active and inactive interfaces by
sending LACPDUs to each other.
The static LACP mode is called the M:N mode. In this mode, both load
balancing and redundancy can be implemented. In a LAG, M links are active
to forward data and perform load balancing, and other N links are inactive.
The inactive links function as backup links and do not forward data. When
one active link fails, the system selects the link with the highest priority
among backup links to replace the faulty link. The link with the highest
priority becomes active and starts to forward data.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 86

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

On the network shown in Figure 3-8, DeviceA and DeviceB are directly
connected and both of them support LACP. The Eth-Trunk working in static
LACP mode can be configured on the two devices to implement load
balancing and link backup. The static LACP mode is mainly applied to
situations where the bandwidth of M links must be assured and a fault
tolerance mechanism is in place. If an active link fails, the system selects the
backup link with the highest priority as the active link.

Figure 3-8 Eth-Trunk in static LACP mode

● Dynamic LACP mode

LACPDU exchange in static and dynamic LACP modes is the same, but the
processing upon an LACP negotiation failure is different:
– In static LACP mode, an Eth-Trunk becomes Down and cannot forward
data after an LACP negotiation failure.
– In dynamic LACP mode, an Eth-Trunk becomes Down after an LACP
negotiation failure, but its member interfaces inherit VLAN attributes of
the Eth-Trunk and enter the Indep state. The member interfaces can still
forward data at Layer 2.
After a device configured with an Eth-Trunk in dynamic LACP mode receives
LACPDUs from the remote device, the two devices will use LACPDUs to
negotiate link aggregation parameters. After the negotiation, link aggregation
provides the same function as the Eth-Trunk working in static LACP mode.
An Eth-Trunk in dynamic LACP mode is often used to directly connect the
device and server. As shown in Figure 3-9, ServerA needs to obtain the
configuration file from ServerB through DeviceA.
– After ServerA restarts and has no configuration, LACP negotiation fails.
The dynamic LACP mode ensures that ServerA obtains the configuration
file from ServerB through an Eth-Trunk member interface.
– After DeviceA receives LACPDUs from ServerA, it uses LACPDUs to
negotiate link aggregation parameters with ServerA.

Figure 3-9 Eth-Trunk in dynamic LACP mode

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 87

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

Eth-Trunk in dynamic LACP mode can be used only in a scenario where a Huawei device is
interconnected with a server. (Ports on the server must be isolated from each other. For
example, NICs on the server cannot be configured to work in bridge mode. If the ports are not
isolated, loops may occur on the network.) In other scenarios, you are advised to deploy Eth-
Trunk in static LACP mode.

3.2.5 Load Balancing Using Link Aggregation

Load balancing can be implemented on a data flow or multiple data flows to
relieve the pressure on a single physical link. A data flow is a group of data
packets with one or more identical attributes. The attributes include the source/
destination MAC address, source/destination IP address, and source/destination
TCP/UDP port number.

When an Eth-Trunk is used to forward data frames, data frames of the same data
flow may be transmitted over different physical links. This mode ensures optimal
bandwidth utilization. However, data frames may arrive at the destination in a
different order to which they were transmitted, resulting to mis-sequencing.

Flow-based load balancing is introduced to prevent this problem. Flow-based load

balancing ensures that frames of the same data flow are forwarded on the same
physical link and implements load balancing of flows. The system achieves this by
using the hash algorithm to calculate the address in a data frame and generate a
HASH-KEY value. Then the system searches for the outbound interface in the Eth-
Trunk forwarding table based on the generated HASH-KEY value. Each MAC or IP
address corresponds to a HASH-KEY value, so the system uses different outbound
interfaces to forward data. Flow-based load balancing ensures the sequence of
data transmission, but cannot ensure the bandwidth utilization.

3.2.6 Preferential Forwarding of Local Traffic on an Eth-Trunk

in a Stack

Concepts
● Stack device
The stack device is a logical device formed by connecting multiple devices
through dedicated stack cables. In Figure 3-10, DeviceB and DeviceC are
connected to form a logical device.
● Inter-device Eth-Trunk
An inter-chassis Eth-Trunk contains physical interfaces of multiple devices in a
stack. When a device in the stack fails or a physical interface added to the
Eth-Trunk fails, traffic can be transmitted between devices through stack
cables. This ensures reliable transmission and implements device backup.
● Preferential forwarding of local traffic
In network b of Figure 3-10, traffic from DeviceB or DeviceC is only
forwarded through local member interfaces when the network runs properly.
In network a of Figure 3-10, traffic is forwarded across devices through stack
cables.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 88

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Figure 3-10 Inter-device Eth-Trunk

Inter-Device Eth-Trunk Supporting Preferential Forwarding of Local Traffic

Inter-chassis Eth-Trunk supporting preferential forwarding of local traffic saves
bandwidth resources between devices and improves traffic forwarding efficiency.
In a stack, an Eth-Trunk is configured to be the outbound interface of traffic to
ensure reliable transmission. Eth-Trunk member interfaces are located on different
devices. If preferential forwarding of local traffic is not enabled, when the stack
device forwards traffic, the Eth-Trunk may select inter-device member interfaces
based on the hash algorithm. This forwarding mode occupies bandwidth resources
between devices and reduces traffic forwarding efficiency.
As shown in Figure 3-10, DeviceB and DeviceC constitute a stack, and the stack
connects to DeviceA through an Eth-Trunk. After the Eth-Trunk in the stack is
configured to preferentially forward local traffic, the following functions are
implemented:
● Forwarding received traffic by the local device
When DeviceB has Eth-Trunk member interfaces and the member interfaces
function properly, the Eth-Trunk forwarding table of DeviceB contains only
local member interfaces. In this manner, the hash algorithm selects a local
member interface, and traffic is only forwarded through DeviceB.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 89

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Forwarding received traffic by another device

When DeviceB does not have any Eth-Trunk member interface or all member
interfaces are faulty, the Eth-Trunk forwarding table of DeviceB contains all
available member interfaces. In this manner, the hash algorithm selects a
member interface on DeviceC, and traffic is forwarded through DeviceC.
NOTE

● This function is only valid for known unicast packets, and is invalid for unknown unicast,
broadcast, and multicast packets.
● Before configuring an Eth-Trunk to preferentially forward local traffic, ensure that
member interfaces of the local Eth-Trunk have sufficient bandwidth to forward local
traffic; otherwise, traffic may be discarded.

3.2.7 Inter-Device Link Aggregation

For link aggregation between standalone devices, if the Eth-Trunk or the remote
device fails, the switch or server cannot communicate with the remote device. This
problem can be resolved by deploying inter-device link aggregation, which allows
a switch or server to be dual-homed to two devices, thereby achieving device-level
link reliability.

Based on the networking of uplink devices, CloudEngine series switches support

three inter-device link aggregation technologies: stacking, M-LAG, and M-LAG Lite.

Stacking
Interfaces on member switches in a stack can be bundled into an Eth-Trunk.

This inter-device link aggregation mode is applied to scenarios where the stack is
connected to other devices, and protects the link between upstream and
downstream devices. The Eth-Trunk can still work even if a member switch fails or
one link of the Eth-Trunk fails, ensuring reliable transmission of data traffic. This
prevents single-point failures of a member device in a stack and greatly improves
the network-wide reliability.

For details, see Stack Configuration in the CloudEngine 9800, 8800, 6800, 5800
Series Switches Configuration Guide - Virtualization Configuration Guide.

Figure 3-11 Physical and logical topologies of inter-device link aggregation in a

stack

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 90

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

M-LAG
M-LAG allows two access switches in the same state to perform link aggregation
negotiation with the access device. In Figure 3-12, the access device
communicates with the M-LAG through link aggregation, achieving device-level
reliability instead of only card-level reliability. The M-LAG is a dual-active system
that is composed of two access switches.
M-LAG is a horizontal virtualization technology that virtualizes two M-LAG devices
into one logical device, that is, a unified Layer 2 logical node. In practice, M-LAG
provides loop-free networking between aggregation and access layers and
replaces STP. Compared with STP, M-LAG provides clear logical topology and
better link use efficiency.
M-LAG master and backup devices forward data simultaneously and their
forwarding behaviors are the same. The forwarding behaviors of M-LAG master
and backup device roles are different only in scenarios where faults occur.
For details, see M-LAG Configuration in the CloudEngine 9800, 8800, 6800, 5800
Series Switches Configuration Guide - Ethernet Switching Configuration Guide.

Figure 3-12 Physical and logical topologies of inter-device link aggregation in an

M-LAG system

M-LAG Lite
Figure 3 shows the inter-device link aggregation on the M-LAG Lite networking.
Compared with inter-device link aggregation in a stack or M-LAG system, inter-
device link aggregation in an M-LAG Lite system removes the heartbeat cable
(iStack link or peer-link) required for establishing a stack/M-LAG between devices.
Inter-device link aggregation in an M-LAG Lite system simplifies deployment,
saves costs, and ensures that services are not interrupted during the upgrade.
In Figure 3, the same Eth-Trunk ID, LACP system ID, LACP system priority, and
different numbers of Eth-Trunk member interfaces are configured on DeviceB and
DeviceC. (On one member device, the lacp port-id-extension enable command is
run in views of all the member interfaces to increase the number of each member
interface by 32768.) In this way, the inter-device Eth-Trunk interface negotiation
can be successful. The two devices evenly load balance data. When one device
fails, traffic can be forwarded through the other device, implementing device-level
reliability.
DeviceB and DeviceC must be Layer 3 gateways and cannot be Layer 2 transparent
transmission devices. If an upstream device functions as the gateway, the ARP and

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 91

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

ND entries of the server or switch learned by the gateway have two egresses,
causing MAC address flapping. Therefore, the devices to which the server or switch
connects in M-LAG Lite mode must be Layer 3 gateways.
For details about M-LAG Lite, see "Example for Configuring M-LAG Lite" in
Configuration Examples for Comprehensive Scenarios in the CloudEngine 9800,
8800, 6800, 5800 Series Switches Configuration Guide - Typical Configuration
Examples. For details about IPv6 M-LAG Lite, see "Example for Configuring IPv6
M-LAG Lite" in Configuration Examples for Comprehensive Scenarios in the
CloudEngine 9800, 8800, 6800, 5800 Series Switches Configuration Guide - Typical
Configuration Examples.

Figure 3-13 Inter-device link aggregation in M-LAG Lite networking

3.3 Application Scenarios for Link Aggregation

3.3.1 Using Eth-Trunk to Connect Two Access Switches to a

Core Switch
As shown in Figure 3-14, access switches SwitchB and SwitchC of the data center
connect to core switch SwitchA. SwitchB and SwitchC connect to many users, and
SwitchA connects to the external network through the egress router. More and
more users connecting to the network want to communicate with each other. The
links between SwitchA and SwitchB, and between SwitchA and SwitchC require
sufficient bandwidth and reliability. Eth-Trunk 1 and Eth-Trunk 2 are then created
to provide high bandwidth and reliability.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 92

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Figure 3-14 Eth-Trunk networking

You can determine the working mode for the Eth-Trunk according to the following
situations:
● If devices at both ends of the Eth-Trunk support LACP, the LACP mode is
recommended.
● If the device at either end of the Eth-Trunk does not support LACP, you must
use the manual load balancing mode.

3.4 Summary of Link Aggregation Configuration Tasks

The device supports link aggregation in the manual load balancing mode and Link
Aggregation Control Protocol (LACP) mode. In a stack, local traffic is preferentially
forwarded by the Eth-Trunk.
Table 3-1 lists the link aggregation configuration tasks.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 93

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Table 3-1 Link aggregation configuration tasks

Item Description Task

Configure link In manual load 3.7 Configuring Link

aggregation in manual balancing mode, you Aggregation in Manual
load balancing mode. must manually create an Load Balancing Mode
Eth-Trunk and add
member interfaces to the
Eth-Trunk. All active links
forward data and evenly
load balance traffic. The
manual load balancing
mode is often used when
the remote device does
not support LACP.

Configure link In LACP mode, you must 3.8 Configuring Link

aggregation in LACP manually create an Eth- Aggregation in LACP
mode. Trunk and add interfaces Mode
to the Eth-Trunk. LACP
determines active
interfaces by negotiating
parameters in LACPDUs.
LACP provides backup
links and ensures high
reliability of member
links.

Configure preferential On a network where the 3.9 Configuring

forwarding of local stack and Eth-Trunk are Preferential Forwarding
traffic in a stack. used, configure the Eth- of Local Traffic on an
Trunk to preferentially Eth-Trunk in a Stack
forward local traffic to
increase bandwidth use
efficiency between stack
devices and improve
traffic forwarding
efficiency.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 94

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Item Description Task

Create an Eth-Trunk Layer 3 Sub-interfaces 3.10 Creating an Eth-

Layer 3 sub-interface. can be configured on a Trunk Layer 3 Sub-
Layer 3 Eth-Trunk. When interface
Layer 3 devices connect
to Layer 2 devices in
different VLANs through
the Layer 3 Eth-Trunk,
Layer 3 sub-interfaces
must be configured on
the Eth-Trunk to identify
packets from different
VLANs and to enable
users in these VLANs to
communicate with each
other.

3.5 Licensing Requirements and Limitations for

Ethernet Link Aggregation
Involved Network Element
Other network elements are not required.

Licensing Requirements
Ethernet link aggregation is a basic function of the switch, and as such is
controlled by the license for basic software functions. The license for basic
software functions has been loaded and activated before delivery. You do not need
to manually activate it.

Version Requirements

Table 3-2 Products and minimum version supporting Ethernet link aggregation

Product Minimum Version Required

CE9860EI V200R020C00

CE8861EI/CE8868EI V200R005C10

CE8850-32CQ-EI V200R002C50

CE8850-64CQ-EI V200R005C00

CE6856HI V200R002C50

CE6857EI V200R005C10

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 95

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Product Minimum Version Required

CE6865EI V200R005C00

CE6870-48S6CQ-EI V200R001C00

CE6870-48T6CQ-EI V200R002C50

CE6881, CE6820, CE6863 V200R005C20

CE6881K V200R019C10

CE6881E V200R019C10

CE6863K V200R019C10

CE5881 V200R020C00

CE5855E V200R020C10

CE6857E V200R020C10

NOTE

For details about the mapping between software versions and switch models, see the
Hardware Query Tool.
Software version evolution:
● For switches excluding the CE9860EI, CE6881, CE6881E, CE6881K, CE6863, CE6863K,
CE6857E, CE6820, CE5881, and CE5855E
V100R001C00 -> V100R002C00 -> V100R003C00 -> V100R003C10 -> V100R005C00 ->
V100R005C10 -> V100R006C00 -> V200R001C00 -> V200R002C50 -> V200R003C00 ->
V200R005C00 -> V200R005C10 -> V200R019C00 -> V200R019C10 -> V200R020C10
● For the CE9860EI, CE6881, CE6881E, CE6881K, CE6863, CE6863K, CE6857E, CE6820,
CE5881, CE5855E
V200R005C20 -> V200R019C10 -> V200R020C00 -> V200R020C10

Feature Limitations
Licensing Requirements and Limitations Before an Eth-Trunk Is Configured
● Each Eth-Trunk supports a maximum of 128 member interfaces on the
CE6881, CE5881, CE6881K, CE6820, CE6863K, CE6881E, and CE6863. Each Eth-
Trunk supports a maximum of 64 member interfaces on the CE9860EI. The
maximum number of member interfaces in each Eth-Trunk on other models is
determined by the assign forward eth-trunk mode command.
● In distributed forwarding mode, switches can be configured with less than 256
link aggregation groups.
● In versions earlier than V200R002C50, member interfaces of an Eth-Trunk
must use the same rate.
For example, GE and 10GE interfaces cannot join the same Eth-Trunk,
whereas GE electrical and optical interfaces can join the same Eth-Trunk.
In V200R002C50 and later versions, Ethernet interfaces working with different
rates can join the same Eth-Trunk.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 96

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● The number of member interfaces of an Eth-Trunk cannot exceed the

maximum number of member interfaces allowed.
● Member interfaces cannot be configured with some services or static MAC
address entries. For example, when an interface is added to an Eth-Trunk, the
interface must use the default link type.
● Member interfaces of an Eth-Trunk cannot be an Eth-Trunk.
● Device connection
– If an interface of the local device is added to an Eth-Trunk, an interface of
the remote device directly connected to the interface of the local device
must also be added to an Eth-Trunk. Otherwise, communication between
the two devices will fail.
– Both devices of the Eth-Trunk must use the same number of physical
interfaces, interface rate, duplex mode, jumbo, and flow control mode.
– Both devices of an Eth-Trunk must use the same link aggregation mode.
● Preferential forwarding of local traffic is only valid for known unicast packets.
It is invalid for broadcast packets, multicast packets, and unknown unicast
packets.
● VLANs, VXLANs, carrier VLANs, and main interfaces share system resources. If
system resources are insufficient, the configurations of these features may fail.
● GRE and IP-in-IP remote mirroring packets cannot be load balanced based on
inner or outer headers.
Licensing Requirements and Limitations After an Eth-Trunk Is Configured
● An Ethernet interface can be added to only one Eth-Trunk. To add an Ethernet
interface to another Eth-Trunk, delete it from the original one first.
● When an Eth-Trunk member interface is removed from an Eth-Trunk or a
physical interface is added to an Eth-Trunk, run the shutdown command for
the Eth-Trunk member interface or physical interface, and then run the undo
shutdown command after the Eth-Trunk member interface is removed from
or the physical interface is added to the Eth-Trunk.
● After an interface is added to an Eth-Trunk, the Eth-Trunk learns MAC address
entries or ARP entries, but the member interface does not.
● Before deleting an Eth-Trunk, delete member interfaces from the Eth-Trunk.
● For CE6870EI, The following services are in descending order of priority: M-
LAG unidirectional isolation, MQC (traffic policing, traffic statistics collection,
and packet filtering), querying the outbound interface of packets with
specified 5-tuple information, source MAC address, and destination MAC
address, local VLAN mirroring, sFlow, NetStream, and VLANIF interface
statistics collection. When the services are configured on an interface in the
outbound direction, only the service with the highest priority takes effect. For
example, when both packet filtering and VLANIF interface statistics collection
are configured on a VLANIF interface, packet filtering takes effect.
For sFlow and NetStream, the preceding limitations apply only to Layer 2 sub-
interfaces and Layer 3 sub-interfaces. For details about the priorities between
MQC-based traffic statistics collection and traffic statistics collection on a
VLANIF interface, see Licensing Requirements and Limitations for Traffic
Statistics Collection.
● The CE6881, CE5881, CE6881K, CE6820, CE6863, CE6863K, CE6881E, and
CE6870EI do not support forwarding packet capture from member interfaces

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 97

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

in a specified Eth-Trunk. The switches can capture the packets to be sent to

the CPU only on Eth-Trunks, but not on member interfaces of the Eth-Trunks.
● The dynamic load balancing for an LAG and elastic load balancing functions
cannot be applied together.

3.6 Default Settings for Link Aggregation

Table 3-3 Default settings for link aggregation
Parameter Default Setting

Link aggregation mode Manual load balancing mode

Upper threshold for the number of 128 on the CE6881, CE5881, CE6881K,
active member links CE6820, CE6863K, CE6881E, and
CE6863, 64 on the CE9860EI. 32 on the
CE6870EI, and 16 on other models.

Lower threshold for the number of 1

active member links

LACP system priority 32768

LACP interface priority 32768

LACP preemption Disabled

LACP preemption delay 30s

Timeout interval at which LACPDUs 90s

are received

Preferentially forwarding local traffic Enabled

on an Eth-Trunk

3.7 Configuring Link Aggregation in Manual Load

Balancing Mode

3.7.1 (Optional) Setting the Number of LAGs

Context
Typically, the number of LAGs supported by a switch is fixed. However, you can
run the assign forward eth-trunk mode command to flexibly set the number of
LAGs supported by a switch, implementing flexible networking and meeting
diversified service requirements.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 98

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

The number of LAGs supported by the CE9860EI, CE6881, CE5881, CE6881K, CE6820,
CE6863, CE6863K, CE6881E is fixed. The CE6881, CE5881, CE6881K, CE6820, CE6863,
CE6863K, CE6881E do not support a variable number of LAGs.

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run assign forward eth-trunk mode { 64 | 128 | 256 | 512 | 1024 }

The number of LAGs supported by the device is set.

For CE6870EI, the number of LAGs supported by the device can be set using the
assign forward eth-trunk mode { 256 | 512 | 1024 } command.

By default, the CE6870EI support 512 LAGs; other models support 128 LAGs.

NOTE

After the assign forward eth-trunk mode command is used to change the number of
LAGs supported by the device, restart the device to make the configuration take effect.

Step 3 (Optional) Run eth-trunk reserved for ipv6 vxlan number

Configure the number of Eth-Trunk ID resources reserved for the IPv6 VXLAN
function.

NOTE

Only the CE6857EI, CE6857E, CE6865EI, CE8861EI, and CE8868EI support this command.

Step 4 Run commit

The configuration is committed.

----End

3.7.2 Creating a LAG

Context
Each LAG has one logical interface, that is, an Eth-Trunk. Before configuring link
aggregation, create an Eth-Trunk.

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run interface eth-trunk trunk-id

An Eth-Trunk is created and the Eth-Trunk interface view is displayed.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 99

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

If the specified Eth-Trunk already exists, this command directly displays the Eth-
Trunk interface view.
Advised to create Eth-Trunks in ascending order.
trunk-id defines the Eth-Trunk ID. The value ranges of different models are
different. Table 3-4 lists value ranges of different models.

Table 3-4 Value ranges of different models

Product Model Value Range

CE9860EI The value range cannot be adjusted. The value of trunk-id

is in the range 0 to 143. Each Eth-Trunk allows a
maximum of 64 member interfaces.

CE6870EI The value range can be adjusted using the assign forward
eth-trunk mode { 256 | 512 | 1024 } command.
● When 256 is specified, the value is in the range 0 to
255. Each Eth-Trunk allows a maximum of 64 member
interfaces.
● When 512 is specified, the value is in the range 0 to
511. Each Eth-Trunk allows a maximum of 32 member
interfaces. The default value is 512.
● When 1024 is specified, the value is in the range 0 to
1023. Each Eth-Trunk allows a maximum of 16 member
interfaces.

CE6881, CE5881, The value range cannot be adjusted. The value of trunk-id
CE6881K, CE6820, is in the range 0 to 1023. Each Eth-Trunk allows a
CE6863, CE6863K, maximum of 128 member interfaces.
CE6881E

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 100

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Product Model Value Range

CE8850-64CQ-EI The value range can be adjusted using the assign forward
eth-trunk mode { 32 | 64 | 128 | 256 | 512 | 1024 }
command.
● When 32 is specified, the value is in the range 0 to 31.
Each Eth-Trunk allows a maximum of 64 member
interfaces.
● When 64 is specified, the value is in the range 0 to 63.
Each Eth-Trunk allows a maximum of 32 member
interfaces.
● When 128 is specified, the value is in the range 0 to
127. Each Eth-Trunk allows a maximum of 16 member
interfaces. The default value is 128.
● When 256 is specified, the value is in the range 0 to
255. Each Eth-Trunk allows a maximum of 8 member
interfaces.
● When 512 is specified, the value is in the range 0 to
511. Each Eth-Trunk allows a maximum of 4 member
interfaces.
● When 1024 is specified, the value is in the range 0 to
1023. Each Eth-Trunk allows at most 2 member
interfaces.

Other models: The value range can be adjusted using the assign forward
eth-trunk mode { 64 | 128 | 256 | 512 | 1024 } command.
● When 64 is specified, the value is in the range 0 to 63.
Each Eth-Trunk allows a maximum of 32 member
interfaces.
● When 128 is specified, the value is in the range 0 to
127. Each Eth-Trunk allows a maximum of 16 member
interfaces. The default value is 128.
● When 256 is specified, the value is in the range 0 to
255. Each Eth-Trunk allows a maximum of 8 member
interfaces.
● When 512 is specified, the value is in the range 0 to
511. Each Eth-Trunk allows a maximum of 4 member
interfaces.
● When 1024 is specified, the value is in the range 0 to
1023. Each Eth-Trunk allows at most 2 member
interfaces.

Step 3 Run commit

The configuration is committed.

----End

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 101

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3.7.3 Setting the Link Aggregation Mode to Manual Load

Balancing

Context
Link aggregation can work in manual load balancing mode and LACP mode.
In manual load balancing mode, you must manually create an Eth-Trunk and add
member interfaces to the Eth-Trunk. All active links forward data and evenly load
balance traffic. The manual load balancing mode is used when the remote device
does not support LACP.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
Step 3 Run mode manual [ load-balance ]
The Eth-Trunk is configured to work in manual load balancing mode.
By default, an Eth-Trunk works in manual load balancing mode.
Before configuring an Eth-Trunk, ensure that both devices use the same working
mode.
Step 4 Run commit
The configuration is committed.

----End

3.7.4 Adding Member Interfaces to an Eth-Trunk

Context
You can add member interfaces to an Eth-Trunk in the Eth-Trunk interface view or
member interface view.
When adding an interface to an Eth-Trunk, pay attention to the following points:
● On the CE6881, CE5881, CE6881K, CE6820, CE6863K, CE6881E, and CE6863,
an Eth-Trunk contains a maximum of 128 member interfaces. On the
CE9860EI, an Eth-Trunk contains a maximum of 64 member interfaces. On
other models, the number of member interfaces in an Eth-Trunk depends on
the assign forward eth-trunk mode command.
● A member interface cannot be configured with some services or static MAC
addresses.
● When adding an interface to an Eth-Trunk, ensure that the interface uses the
default link type.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 102

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● An Eth-Trunk cannot be added to another Eth-Trunk.

● An Ethernet interface can be added to only one Eth-Trunk. To add the
Ethernet interface to another Eth-Trunk, delete it from the Eth-Trunk first.
● If an interface of the local device is added to an Eth-Trunk, an interface of the
remote device directly connected to the interface of the local device must also
be added to an Eth-Trunk so that the two ends can communicate.
● After interfaces are added to an Eth-Trunk, the Eth-Trunk learns MAC
addresses and ARP entries but member interfaces do not.
● Devices at both ends of an Eth-Trunk must use the same number of physical
interfaces, interface rate, duplex mode, jumbo, and flow control mode.
● In a stack scenario, it is recommended that the number of member interfaces
added to a LAG be the nth power of 2. Otherwise, unknown unicast traffic
may be unevenly load balanced.

Procedure
● Add member interfaces to an Eth-Trunk in the Eth-Trunk interface view.
a. Run system-view
The system view is displayed.
b. Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
c. Run trunkport interface-type { interface-number1 [ to interface-
number2 ] } &<1-n>
A member interface is added to the Eth-Trunk.
When you add member interfaces to an Eth-Trunk in a batch, if one
interface cannot be added to the Eth-Trunk, all subsequent interfaces in
the batch cannot be added to the Eth-Trunk, either.

NOTE

The value of n is 128.

d. Run commit
The configuration is committed.
● Add member interfaces to an Eth-Trunk in the member interface view.
a. Run system-view
The system view is displayed.
b. Run interface interface-type interface-number
The member interface view is displayed.
c. Run eth-trunk trunk-id
The member interface is added to an Eth-Trunk.
d. Run commit
The configuration is committed.
----End

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 103

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Follow-up Procedure
When the status of an Eth-Trunk member interface changes, the system sends
traps containing the status change information for you to confirm whether the
device encounters any fault. If you want to know the ID of the Eth-Trunk to which
the member interface belongs, run the trunk-member trap in private-mib
enable command to enable Eth-Trunk member interfaces to send traps through a
private MIB. Traps sent through a private MIB carry the number of the specific
Eth-Trunk interface.

After the trunk-member trap in private-mib enable command is run, traps are
sent through a private MIB, instead of a public MIB. In this case, you can view the
traps only by using the private MIB of Huawei.

3.7.5 (Optional) Setting the Lower Threshold for the Number

of Active Interfaces

Context
The lower threshold for the number of active interfaces affects the status and
bandwidth of an Eth-Trunk. To ensure that the Eth-Trunk functions properly and is
less affected by member link status changes, set the lower threshold for the
number of active interfaces.

When the number of active interfaces falls below the lower threshold, the Eth-
Trunk goes Down. This ensures that the Eth-Trunk has a minimum available
bandwidth.

The upper threshold for the number of active interfaces is inapplicable to the
manual load balancing mode.

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run interface eth-trunk trunk-id

The Eth-Trunk interface view is displayed.

Step 3 Run least active-linknumber link-number

The lower threshold for the number of active interfaces is set.

By default, the lower threshold for the number of active interfaces is 1.

The lower threshold for the number of active interfaces on the local switch can be
different from that on the remote switch.

Step 4 Run commit

The configuration is committed.

----End

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 104

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3.7.6 (Optional) Configuring the Weight of Load Balancing for

a Member Interface

Context
On an Eth-Trunk interface, you can load balance traffic among member interfaces
according to the weights configured for the member interfaces.

The higher the weight of a member interface, the heavier the load over the
member link. Therefore, you can configure a higher weight for a member interface
so that the member link can carry a heavier load.

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run interface interface-type interface-number

The Eth-Trunk member interface view is displayed.

Step 3 Run distribute-weight weight-value

The load balancing weight is configured for the Eth-Trunk member interface.

The default weight of an Eth-Trunk member interface is 1.

The total weight of member interfaces in an Eth-Trunk cannot exceed the

maximum number of member interfaces allowed.

Step 4 Run commit

The configuration is committed.

----End

3.7.7 (Optional) Configuring a Load Balancing Mode (CE

Switches Excluding the CE6870EI)

Context
An Eth-Trunk supports per-packet load balancing and per-flow load balancing.
● Per-packet load balancing can improve Eth-Trunk bandwidth efficiency to
ensure even load balancing among equal-cost routes, but cannot prevent
packet mis-sequencing. To ensure packet sequencing, confirm that the device
or terminal receiving traffic supports packet reassembly in case of packet mis-
sequencing. Switches support the following per-packet load balancing modes:
– Random mode: The outbound interface of packets is generated randomly
and calculated based on the time when the packets reach the Eth-Trunk.
When the IP address and MAC address of known unicast packets remain
unchanged, configure random per-packet load balancing.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 105

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

– Round-robin mode: Eth-Trunk member interfaces forward traffic in turn.

When known unicast packets have a similar length, configure round-
robin per-packet load balancing.
● Per-flow load balancing ensures that packets of the same data flow are
forwarded on the same physical link and those of different data flows are
forwarded on different physical links. Table 3-5 lists the load balancing
modes for different types of packets.

Table 3-5 Load balancing modes for different types of packets

Packet Default Load Configurable Remarks
(Inbound Balancing Mode Load Balancing
Interface) Mode

IPv4 packets src-ip, dst-ip, l4- src-ip, dst-ip, l4- The load
src-port and l4- src-port, l4-dst- balancing mode
dst-port port, and is relevant to the
protocol packet type and
irrelevant to the
IPv6 packets src-ip, dst-ip, l4- src-ip, dst-ip, packet
src-port, and l4- protocol, l4-src- forwarding
dst-port port, and l4-dst- process.
port
For example,
MPLS packets Ingress/Egress/ For switches even if the
Transit: top- excluding the system provides
label and 2nd- CE6865EI, only Layer 2
label CE6857EI, forwarding for
CE6857E, IPv4 packets, the
CE8861EI, IPv4 packets are
CE8868EI: load balanced
according to the
Ingress/Egress/
load balancing
Transit: top-
mode for IPv4
label, 2nd-label,
packets. When
dst-ip, and src-ip
the system
For the cannot identify
CE6857EI, IPv4, IPv6, or
CE6857E, MPLS packets,
CE6865EI, the system load
CE8861EI, and balances packets
CE8868EI: based on src-
Ingress/Egress/ mac, dst-mac,
Transit: top- src-interface,
label, 2nd-label, and eth-type for
3rd-label, 4th- Layer 2 packets.
label, 5th-label,
dst-ip, src-ip,
and src-
interface

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 106

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Packet Default Load Configurable Remarks

(Inbound Balancing Mode Load Balancing
Interface) Mode

Layer 2 packets src-mac and dst- src-mac, dst-

except IPv4, IPv6, mac mac, src-
and MPLS interface, and
packets eth-type

TRILL packets Ingress node: src-mac, dst- TRILL packets

inner src-mac mac, src-ip, dst- can be load
and dst-mac for ip, src-interface, balanced on the
Layer 2 packets; l4-src-port, l4- transit node only
src-ip, dst-ip, l4- dst-port when the load-
src-port, and l4- balance
dst-port for enhanced
Layer 3 packets [ resilient ]
profile profile-
Transit/Egress src-mac, dst- name command
node: inner src- mac, src-ip, dst- is used.
mac and dst- ip, l4-src-port
mac for Layer 2 and l4-dst-port
packets; src-ip,
dst-ip, l4-src-
port, and l4-dst-
port for Layer 3
packets

FCoE packets dst-fcid and src- dst-fcid and src- -

fcid fcid

NOTE

Load balancing is valid only for outgoing traffic; therefore, the load balancing modes for the
interfaces at both ends of the link can be different and do not affect each other.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 (Optional) Run load-balance profile profile-name
A load balancing profile is configured and its view is displayed. profile-name
specifies the name of the load balancing profile.
By default, there is a load balancing profile named default.
Run the following commands as required. You can configure a load balancing
mode for Layer 2 packets, IPv4 packets, IPv6 packets, FCoE, and MPLS packets
respectively.

● Run l2 [ src-mac | dst-mac | src-interface | eth-type ] *

A load balancing mode is configured for Layer 2 packets (non-IP packets).

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 107

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

By default, the switch load balances Layer 2 packets (non-IP packets) based
on the source MAC address (src-mac) and destination MAC address (dst-
mac).
● Run ip [ src-ip | dst-ip | l4-src-port | l4-dst-port | protocol ] *

A load balancing mode is configured for IPv4 packets.

By default, the switch load balances IPv4 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
● Run ipv6 [ src-ip | dst-ip | protocol | l4-src-port | l4-dst-port ] *

A load balancing mode is configured for IPv6 packets.

By default, the switch load balances IPv6 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
● Run mpls [ top-label | dst-ip | src-ip | 2nd-label | 3rd-label | 4th-label | 5th-
label | src-interface ] *
A load balancing mode is configured for MPLS packets.
By default, the switch load balances MPLS packets based on the two outer
labels (top-label and 2nd-label).
NOTE

Only the CE6857EI, CE6857E, CE6865EI, CE8861EI, and CE8868EI support 3rd-label,
4th-label, 5th-label, and src-interface. If src-ip or dst-ip is configured, 4th-label or
5th-label cannot be configured.
In V200R005C10 and earlier versions, if 4th-label and 5th-label are configured for
load balancing of MPLS packets in the load balancing profile view or ECMP view, this
configuration takes effect for both Eth-Trunk and ECMP. The load balancing mode
based on src-ip and dst-ip conflicts with that based on 4th-label and 5th-label, and
the last delivered configuration takes effect. Therefore, the effective load balancing
mode is inconsistent with that in the configuration file.
● Run fcoe { dst-fcid | src-fcid } *

A load balancing mode is configured for FCoE packets.

By default, the switch load balances FCoE packets based on the source FC_ID
(src-fcid) and destination FC_ID (dst-fcid).
● Run tunnel {outer-header | inner-header}
Load balancing based on the inner or outer header is configured for tunnel
packets (GRE and IP-in-IP packets) in the specified load balancing profile.
By default, GRE and IP-in-IP packets are load balanced based on inner
headers.
NOTE

This command is supported on the following models: CE5855E, CE6856HI, CE6857EI,

CE6857E, CE6865EI, CE8850EI, CE8861EI, CE8868EI, CE9860EI.
● Run eth-trunk { hash-mode hash-mode-id | universal-id universal-id } *

A load balancing mode is configured for the Eth-Trunk.

By default, an Eth-Trunk load balances packets based on hash-mode (1) and
universal-id (1).

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 108

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Run stack { hash-mode hash-mode-id | universal-id universal-id } *

A load balancing mode is configured for a stack port or fabric port in the
specified load balancing profile.
By default, a stack port or fabric port load balances packets based on hash-
mode (1) and universal-id (1) in a load balancing profile.

Step 3 (Optional) Run quit

Return to the system view.

Step 4 Run the following commands as required.

● Configure a load balancing mode for known unicast traffic.
a. Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
b. Run load-balance { dst-ip | dst-mac | random | round-robin | src-ip |
src-mac | src-dst-ip | src-dst-mac | enhanced [ resilient ] profile profile-
name }
A load balancing mode is configured for the Eth-Trunk.
By default, the load balancing mode of the Eth-Trunk is enhanced
profile.
NOTE

Only the CE8868EI, CE8861EI, CE6881K, CE6881E, CE6881, CE5881, CE6865EI,

CE6863K, CE6863, CE6857EI, CE6857E, CE6856HI , CE6820 support resilient
parameter.
Only the CE8868EI, CE8861EI, CE8850EI, CE6865EI, and CE6857EI, CE6857E support
random and round-robin parameters.
CE9860EI only supports random and enhanced parameters.

▪ dst-ip: based on destination IP addresses

▪ dst-mac: based on destination MAC addresses

▪ random: The outbound interface of packets is generated randomly

and calculated based on the time when the packets reach the Eth-
Trunk. When the IP address and MAC address of known unicast
packets do not change, use this mode to perform packet-based load
balancing so that traffic is load balanced. The random mode causes
the mis-sequencing problem. Ensure that the receive device or
terminal supports assembly of mis-sequenced packets.

▪ round-robin: Each Eth-Trunk member interface forwards traffic in

turn. For known unicast traffic, if packets have approximate lengths,
configure this load balancing mode to achieve even load balancing.
This mode may cause the mis-sequencing problem. Ensure that the
receive device or terminal supports assembly of mis-sequenced
packets.

▪ src-ip: based on source IP addresses

▪ src-mac: based on source MAC addresses

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 109

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

▪ src-dst-ip: based on the Exclusive-Or result of source and destination

IP addresses

▪ src-dst-mac: based on the Exclusive-Or result of source and

destination MAC addresses

▪ enhanced [ resilient ] profile: based on the fields in the global load

balancing profile. resilient indicates that traffic on links is switched
less as much as possible when links are increased or reduced and
only some traffic is switched between links. For example, an LAG has
three member links, data is forwarded based on the hash key, and
one link is faulty. When resilient is not specified, traffic is reallocated
on the other two links. When resilient is specified, traffic that has
been allocated on the other two links remain unchanged, and traffic
on the faulty link is evenly allocated on the two links. Therefore,
there is less impact on services. When the faulty link recovers, some
traffic on the other two links is switched to it. Traffic allocation on
links is different from that before the fault occurrence.
c. Run commit
The configuration is committed.
● Configure a load balancing mode for unknown unicast traffic.
a. Run load-balance unknown-unicast { mac | enhanced }
A load balancing mode is configured.
By default, the load balancing mode is enhanced.
b. Run commit
The configuration is committed.

----End

Follow-up Procedure
● Configure simulated calculation of an Eth-Trunk outbound interface.
Configure simulated calculation of an Eth-Trunk outbound interface after
specifying the 5-tuple information, source MAC address, and destination
address.
display load-balance forwarding-path unicast interface eth-trunk trunk-id src-interface interface-
type interface-number { ethtype ethtype-number | vlan vlan-id | [ [ src-ip src-ip-data | dst-ip dst-ip-
data ] * | [ src-ipv6 src-ipv6-data | dst-ipv6 dst-ipv6-data ] * ] | src-mac src-mac-data | dst-mac dst-
mac-data | protocol { protocol-number | icmp | igmp | ip | ospf | tcp [ l4-src-port src-port-data | l4-
dst-port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-dst-port dst-port-data ] * } } * slot slot-
id
● Verifying the Configuration
Verify the outbound interface of packets that contain specified 5-tuple
information, source MAC address, and destination MAC address.
display port forwarding-path { src-ip src-ip-data | dst-ip dst-ip-data | src-mac src-mac-data | dst-
mac dst-mac-data | protocol { protocol-number | gre | icmp | igmp | ip | ipinip | ospf | tcp [ l4-src-
port src-port-data | l4-dst-port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-dst-port dst-
port-data ] * } } *

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 110

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3.7.8 (Optional) Configuring a Load Balancing Mode

(CE6870EI)

Context
An Eth-Trunk uses flow-based load balancing. Per-flow load balancing ensures
that packets of the same data flow are forwarded on the same physical link and
those of different data flows are forwarded on different physical links.
Load balancing is valid only for outgoing traffic; therefore, interfaces at both ends
of the link can use different load balancing modes.
Table 3-6 lists load balancing modes for different types of packets.

Table 3-6 Load balancing modes for different types of packets

Packet (Inbound Default Load Balancing Configurable Load
Interface) Mode Balancing Mode

IPv4 packets src-ip, dst-ip, l4-src- src-ip, dst-ip, l4-src-

port, and l4-dst-port port, l4-dst-port, and
protocol

IPv6 packets src-ip, dst-ip, l4-src- src-ip, dst-ip, l4-src-

port, and l4-dst-port port, l4-dst-port, flow-
label, and protocol
NOTE
IPv6 packet load balancing
modes, l4-src-port and l4-
dst-port, are affected by
the l4-src-port and l4-dst-
port fields of IPv4 packets.
That is, when the load
balancing modes of IPv4
packets include l4-src-port
or l4-dst-port, the l4-src-
port or l4-dst-port field
also participates in load
balancing of IPv6 packets.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 111

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Packet (Inbound Default Load Balancing Configurable Load

Interface) Mode Balancing Mode

MPLS packets Ingress/Egress/Transit: Ingress/Egress/Transit:

top-label, 2nd-label, top-label, 2nd-label,
fields in the inner IP and 3rd-label
header (IPv4: src-ip, dst- NOTE
ip, l4-src-port, and l4- The load balancing mode
dst-port; IPv6: src-ip, based on fields in the inner
dst-ip, l4-src-port, and IP header (IPv4: src-ip, dst-
ip, l4-src-port, and l4-dst-
l4-dst-port), and 2nd-
port; IPv6: src-ip, dst-ip,
label l4-src-port, and l4-dst-
port) does not need to be
configured, but the fields
that participate in load
balancing are affected by
the load balancing mode
configuration of IPv4 and
IPv6 packets. If the l4-src-
port and l4-dst-port fields
are specified in the
configured load balancing
mode, packets may be
unable to be load balanced
in the configured load
balancing mode. To
prevent this problem,
cancel the configuration of
the l4-src-port and l4-dst-
port fields in the load
balancing mode of IPv4
and MPLS packets.

VPLS packets Ingress/Egress/Transit: Transit node: inner-sip

top-label, fields in the and inner-dip
inner ETH header (src- PE node: inner-ip and
mac, dst-mac, and ETH inner-ipv6
TYPE), and 2nd-label
NOTE
For VPLS packets, the load
balancing mode based on
fields in the inner Ethernet
header (src-mac, dst-mac
and eth-type) does not
need to be configured, but
the fields that participate
in load balancing are
affected by the load
balancing mode
configuration of the l2
[ src-mac | dst-mac | vlan
| eth-type ] * command.

Layer 2 packets except src-mac, dst-mac and src-mac, dst-mac, vlan,

IPv4, IPv6, and MPLS vlan and eth-type
packets

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 112

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Packet (Inbound Default Load Balancing Configurable Load

Interface) Mode Balancing Mode

TRILL packets Ingress node: src-mac, src-mac, dst-mac, src-ip,

dst-mac, and vlan for dst-ip, l4-src-port, l4-
Layer 2 packets; src-ip, dst-port, and protocol
dst-ip, l4-src-port, and
l4-dst-port for Layer 3
packets

Transit node: src-mac, src-mac, dst-mac, vlan,

dst-mac, vlan, src-ip, src-ip, and dst-ip
and dst-ip in the inner
tag

Egress node: src-mac, src-mac, dst-mac, vlan,

dst-mac, vlan, src-ip, src-ip, and dst-ip
and dst-ip in the inner
tag

FCoE packets dst-fcid and src-fcid dst-fcid, src-fcid

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run load-balance profile profile-name
A load balancing profile is configured and its view is displayed. profile-name
specifies the name of the load balancing profile.
By default, there is a load balancing profile named default.
Step 3 Run the following commands as required. You can configure load balancing modes
for Layer 2, IPv4, IPv6, MPLS, VPLS, TRILL, and FCoE packets, respectively.
● Run l2 [ src-mac | dst-mac | vlan | eth-type ] *

A load balancing mode is configured for Layer 2 packets (not IP packets) in

the specified load balancing profile.
NOTE

During Layer 2 forwarding, the switch load balances only IPv4, IPv6, 802.1ah, ARP, and
CFM packets based on eth-type.
During Layer 2 forwarding, an Eth-Trunk configured with Dot1q tunnel, VLAN stacking,
VLL, or VPLS cannot load balance packets based on the VLAN ID.
● Run ip [ src-ip | dst-ip | l4-src-port | l4-dst-port | protocol ] *

A load balancing mode is configured for IPv4 packets.

By default, the switch load balances IPv4 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 113

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Run ipv6 [ src-ip | dst-ip | protocol | flow-label ] *

A load balancing mode is configured for IPv6 packets.

By default, the switch load balances IPv6 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
NOTE

To load balance IPv6 packets based on the transport-layer source and destination port
numbers, run the ip [ l4-src-port | l4-dst-port ] * command.

● Run mpls [ 2nd-label | 3rd-label | top-label ] * or mpls [ src-ip | dst-ip | l4-

src-port | l4-dst-port ] *
A load balancing mode is configured for MPLS packets in the specified load
balancing profile.
By default, MPLS packets are load balanced based on top-label, 2nd-label,
and fields in the inner IP header (IPv4/IPv6: src-ip, dst-ip, l4-src-port, and l4-
dst-port).
NOTE

On an IPv6 underlay network, if the IPv6 VXLAN function is enabled and the switch
functions as a transit or egress node on the MPLS network, configurations related to
the hash field in inner information in MPLS packets do not take effect in the load
balancing profile. In this scenario, the mpls inner-type { ipv4 | ipv6 | l2 | any }
command cannot be used to change the load balancing mode for MPLS packets in the
load balancing profile. Instead, you can use the source and destination IP addresses in
inner information in MPLS packets to configure a load balancing mode in the load
balancing profile.
– The mpls [ inner-ip | inner-ipv6 ] * command can be configured only when the
IPv6 VXLAN function is enabled. In addition, this command cannot be used to
configure a load balancing mode for VPLS packets. If the IPv6 VXLAN function is
disabled, the switch automatically deletes the configurations related to the mpls
[ inner-ip | inner-ipv6 ] * command.
– For the CE6870EI, if MPLS packets with four or more labels are received, Eth-Trunk
or ECMP load balancing cannot be implemented based on the inner IP header.
– For the CE6870EI, if the implicit null label is disabled and the outer label is popped
out (for example, ASBR in inter-AS VPN Option B and inter-AS VPN Option C
networking), packets may be unable to be load balanced in the load balancing
mode using ECMP or Eth-Trunk based on the inner field in MPLS packets.
● Run user-defined ethernet-over-mpls [ dot1q-tagtag-number ] { inner-
sipsip-offsetsip-nybble-number | inner-dipdip-offsetdip-nybble-number } *
Load balancing based on the inner IP address of VPLS packets is configured
on the outbound Eth-Trunk on the transit node.
● Run user-defined ethernet-over-mpls { inner-ip | inner-ipv6 }
Load balancing based on the inner IP address of VPLS packets is configured
on the outbound Eth-Trunk on the PE node.
NOTE

If inner-ip is specified, you need to run ip [ src-ip | dst-ip ] *. If inner-ipv6 is specified,

you need to run ipv6 [ src-ip | dst-ip ] *. Otherwise, the user-defined ethernet-over-
mpls { inner-ip | inner-ipv6 } command does not take effect.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 114

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Run fcoe { dst-fcid | src-fcid } *

A load balancing mode is configured for FCoE packets.

By default, the switch load balances FCoE packets based on the source FC_ID
(src-fcid) and destination FC_ID (dst-fcid).
● Run trill egress mode { l2 | ipv4 | ipv6 }
An Eth-Trunk load balancing mode is configured on egress and transit nodes
of the TRILL network.
By default, the Eth-Trunk load balancing mode on egress and transit nodes of
the TRILL network is ipv4.
– l2: indicates load balancing based on the source MAC address,
destination MAC address, and VLAN ID. The items used in load balancing
depend on the l2 command.
– ipv4: indicates load balancing based on the source MAC address,
destination MAC address, source IPv4 address, destination IPv4 address,
and VLAN ID. The items used in load balancing depend on the l2 and ip
commands.
– ipv6: indicates load balancing based on the source MAC address,
destination MAC address, source IPv6 address, destination IPv6 address,
and VLAN ID. The items used in load balancing depend on the l2 and
ipv6 commands.
● Run tunnel {outer-header | inner-header}*
Load balancing based on the inner or outer header is configured for tunnel
packets (GRE and IP-in-IP packets) in the specified load balancing profile.
By default, GRE and IP-in-IP packets are load balanced based on both inner
and outer headers.
NOTE

Only CE6870EI Supports this command.

● Run ip-tunnel { inner-src-ip | inner-dst-ip | inner-l4-src-port | inner-l4-dst-
port | inner-protocol } *
Load balancing based on the inner header is configured for tunnel packets
(GRE and IP-in-IP packets) in the specified load balancing profile.
NOTE

– This command does not take effect when GRE and IP-in-IP packets are load
balanced based on the outer header.
– Only CE6870EI Supports this command.
● Run eth-trunk { src-interface | seedseed-data | universal-iduniversal-id |
hash-modehash-mode-id } *
An Eth-Trunk load balancing mode is configured in the specified load
balancing profile.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 115

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

Packets forwarded at Layer 3 cannot be load balanced based on destination MAC

addresses, source MAC addresses, VLAN IDs, and Ethernet types.
Eth-Trunk and ECMP on the device share the same load balancing profile. The load
balancing mode configured using the l2, ip, ipv6, or mpls command takes effect for
both link aggregation and ECMP. The load balancing modes configured using the eth-
trunk (load-balance-profile view) and ecmp (load-balance-profile view)
commands take effect only for link aggregation and ECMP respectively.
When load balancing using both Eth-Trunk and ECMP is uneven, change the value of
universal-id or the hash mode.
● Run mode symmetry
A method is configured for an Eth-Trunk to hash a pair of flows to the same
interface.
By default, no method is configured for an Eth-Trunk to hash a pair of flows
to the same interface.

Step 4 Run commit

The configuration is committed.

----End

Follow-up Procedure
● Configure simulated calculation of an Eth-Trunk outbound interface.
Run the display load-balance forwarding-path unicast interface eth-trunk
trunk-id src-interface interface-type interface-number { ethtype ethtype-
number | vlan vlan-id | [ [ src-ip src-ip-data | dst-ip dst-ip-data ] * | [ src-ipv6
src-ipv6-data | dst-ipv6 dst-ipv6-data | flow-label label-data ] * ] | src-mac
src-mac-data | dst-mac dst-mac-data | protocol { protocol-number | icmp |
igmp | ip | ospf | tcp [ l4-src-port src-port-data | l4-dst-port dst-port-data ] *
| udp [ l4-src-port src-port-data | l4-dst-port dst-port-data ] * } } * slot slot-id
command to configure simulated calculation of an Eth-Trunk outbound
interface after the 5-tuple information, source MAC address, and destination
address are specified.
● Verify the configuration.
Run the display port forwarding-path { src-ip src-ip-data | dst-ip dst-ip-data
| src-mac src-mac-data | dst-mac dst-mac-data | protocol { protocol-number |
gre | icmp | igmp | ip | ipinip | ospf | tcp [ l4-src-port src-port-data | l4-dst-
port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-dst-port dst-port-
data ] * } } * [ enhanced ] command to check the outbound interface of
packets that contain specified 5-tuple information, source MAC address, and
destination MAC address.
Run the display port forwarding-path mpls { { src-ip src-ip-data [ ip-mask-
len | source-ip-mask ] | dst-ip dst-ip-data [ ip-mask-len | dst-ip-mask ] } * |
{ src-ipv6 src-ipv6-data [ ipv6-mask-len ] | dst-ipv6 dst-ipv6-data [ ipv6-
mask-len ] } * }{ transit label-number labelnum | ingress | egress label-
number labelnum } command to check the outbound interface of MPLS
packets that contain information such as the inner source IP address,
destination IP address, role, and number of labels.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 116

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3.7.9 (Optional) Configuring an Eth-Trunk Load Balancing

Mode for PPPoE Packets

Context
By default, a CE switch processes received PPPoE packets as common Ethernet
packets. Figure 3-15 shows the format of PPPoE packets. The switch cannot
identify the 5-tuple information in PPPoE packets and performs the hash
algorithm based on the outer Ethernet frame of common Layer 2 packets by
default. Fields in the Layer 2 frames of PPPoE packets are fixed except the source
MAC address, so packets are often unevenly load balanced on an Eth-Trunk. To
improve the load balancing effect, use inner information of PPPoE packets for load
balancing.

Figure 3-15 Format of PPPoE packets

NOTE

The CE6881, CE5881, CE6881K, CE6820, CE6863, CE6863K, CE6881E can identify PPPoE
packets and load balance the PPPoE packets without configuring Eth-Trunk load balancing.

Procedure
● Configure a load balancing mode for PPPoE packets on the CE6870EI.
a. Run system-view

The system view is displayed.

b. Run load-balance eth-trunk pppoe { session-id | l4-src-port { ppp-
address-compression | ppp-protocol-compression | both | none } }

A load balancing mode is configured for PPPoE packets.

By default, the switch load balances PPPoE packets based on the source
MAC address (smac), destination MAC address (dmac), and VLAN ID
(vlan).

You can specify session-id and l4-src-port so that the switch load
balances PPPoE packets based on the session ID and transport-layer
source port of PPPoE packets.
c. Run commit

The configuration is committed.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 117

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Configure a load balancing mode for PPPoE packets on other models

excluding the CE6870EI.
a. Run system-view
The system view is displayed.
b. Run load-balance profile profile-name
The global load balancing profile view is displayed.
By default, the load balancing profile name is default.
c. Run eth-trunk hash-mode 8
The Eth-Trunk load balancing hash algorithm is set to 8.
d. Run quit
The system view is displayed.
e. Run load-balance ecmp
The ECMP view is displayed.
f. Run hashmode 7
The ECMP load balancing hash algorithm is set to 7.
g. Run quit
The system view is displayed.
h. Run load-balance pppoe { session-id | l4-src-port { ppp-address-
compression | ppp-protocol-compression | both | none } }
A load balancing mode is configured for PPPoE packets.
By default, the switch load balances PPPoE packets based on the source
MAC address (smac) and destination MAC address (dmac).
You can specify session-id and l4-src-port so that the switch load
balances PPPoE packets based on the session ID and transport-layer
source port of PPPoE packets.

NOTE

This command is supported only by the following models: CE6856HI, CE6857EI,

CE6857E, CE6865EI, CE9860EI, and CE8800 series switches.
i. Run commit
The configuration is committed.
----End

3.7.10 (Optional) Binding an Eth-Trunk Member Interface to a

VLAN

Context
A server is often equipped with two or more network adapters. When a server
connects to a switch, one network adapter is used for server management and the

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 118

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

other network adapters are used for traffic forwarding. In this case, there is much
idle time on the network adapter used for server management and switch
interfaces, and the bandwidth use efficiency is low. To address this issue, upgrade
the server software so that the network adapter used for server management can
be also used for traffic forwarding. When the switch uses an Eth-Trunk to connect
to the server, one Eth-Trunk member interface can be bound to a VLAN for server
management. In addition, service VLANs can be configured on the Eth-Trunk so
that service traffic is load balanced among all Eth-Trunk member interfaces. The
bandwidth use efficiency is therefore improved.

NOTE

The CE6870EI, CE9860EI does not support this function.

NOTE

● This function applies to the scenario where a switch connects to a server, and it can only be
configured on the member interface connecting to the server's management NIC.
● In an Eth-Trunk, only one member interface can be bound to a VLAN or VLANs and one
member interface can be bound to a maximum of eight VLANs. Member interfaces of
multiple Eth-Trunks can be bound to the same VLAN.
● The switch supports a maximum of 256 bound VLANs. If N Eth-Trunk member interfaces
are bound to M VLANs, the maximum value of N multiplied by M is 256.
● After this command is configured, packets from the bound VLAN can be only forwarded
through the bound Eth-Trunk member interface. Non-unicast traffic on the Eth-Trunk can
be only forwarded through the bound Eth-Trunk member interface.
● When an Eth-Trunk member interface is bound to a VLAN or VLANs, VLAN mapping, VLAN
stacking, or MUX VLAN cannot be configured.
● When an Eth-Trunk member interface is bound to a VLAN or VLANs, M-LAG cannot be
configured.

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run interface interface-type interface-number

The Eth-Trunk member interface view is displayed.

Step 3 Run trunk member binding vlan vlan-id

The Eth-Trunk member interface is bound to a VLAN.

By default, an Eth-Trunk member interface is not bound to a VLAN.

Step 4 Run commit

The configuration is committed.

----End

3.7.11 Verifying the Link Aggregation Configuration

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 119

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Procedure
● Run the display eth-trunk [ trunk-id [ interface interface-type interface-
number | verbose ] | brief ] command to check the Eth-Trunk configuration.
● Run the display eth-trunk membership trunk-id command to check
information about Eth-Trunk member interfaces.
● Run the display load-balance profile [ profile-name ] command to check the
load balancing profile of the Eth-Trunk.
● Run the display forward eth-trunk mode command to check the number of
LAGs supported by the device.
NOTE

The CE9860EI, CE6881, CE5881, CE6881K, CE6820, CE6863, CE6863K, and CE6881E do
not support this command.

----End

3.8 Configuring Link Aggregation in LACP Mode

3.8.1 (Optional) Setting the Number of LAGs

Context
Typically, the number of LAGs supported by a switch is fixed. However, you can
run the assign forward eth-trunk mode command to flexibly set the number of
LAGs supported by a switch, implementing flexible networking and meeting
diversified service requirements.

NOTE

The number of LAGs supported by the CE9860EI, CE6881, CE5881, CE6881K, CE6820,
CE6863, CE6863K, CE6881E is fixed. The CE6881, CE5881, CE6881K, CE6820, CE6863,
CE6863K, CE6881E do not support a variable number of LAGs.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run assign forward eth-trunk mode { 64 | 128 | 256 | 512 | 1024 }
The number of LAGs supported by the device is set.
For CE6870EI, the number of LAGs supported by the device can be set using the
assign forward eth-trunk mode { 256 | 512 | 1024 } command.
By default, the CE6870EI support 512 LAGs; other models support 128 LAGs.

NOTE

After the assign forward eth-trunk mode command is used to change the number of
LAGs supported by the device, restart the device to make the configuration take effect.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 120

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Step 3 (Optional) Run eth-trunk reserved for ipv6 vxlan number

Configure the number of Eth-Trunk ID resources reserved for the IPv6 VXLAN
function.

NOTE

Only the CE6857EI, CE6857E, CE6865EI, CE8861EI, and CE8868EI support this command.

Step 4 Run commit

The configuration is committed.

----End

3.8.2 Creating a LAG

Context
Each LAG has one logical interface, that is, an Eth-Trunk. Before configuring link
aggregation, create an Eth-Trunk.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run interface eth-trunk trunk-id
An Eth-Trunk is created and the Eth-Trunk interface view is displayed.
If the specified Eth-Trunk already exists, this command directly displays the Eth-
Trunk interface view.
Advised to create Eth-Trunks in ascending order.
trunk-id defines the Eth-Trunk ID. The value ranges of different models are
different. Table 3-7 lists value ranges of different models.

Table 3-7 Value ranges of different models

Product Model Value Range

CE9860EI The value range cannot be adjusted. The value of trunk-id

is in the range 0 to 143. Each Eth-Trunk allows a
maximum of 64 member interfaces.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 121

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Product Model Value Range

CE6870EI The value range can be adjusted using the assign forward
eth-trunk mode { 256 | 512 | 1024 } command.
● When 256 is specified, the value is in the range 0 to
255. Each Eth-Trunk allows a maximum of 64 member
interfaces.
● When 512 is specified, the value is in the range 0 to
511. Each Eth-Trunk allows a maximum of 32 member
interfaces. The default value is 512.
● When 1024 is specified, the value is in the range 0 to
1023. Each Eth-Trunk allows a maximum of 16 member
interfaces.

CE6881, CE5881, The value range cannot be adjusted. The value of trunk-id
CE6881K, CE6820, is in the range 0 to 1023. Each Eth-Trunk allows a
CE6863, CE6863K, maximum of 128 member interfaces.
CE6881E

CE8850-64CQ-EI The value range can be adjusted using the assign forward
eth-trunk mode { 32 | 64 | 128 | 256 | 512 | 1024 }
command.
● When 32 is specified, the value is in the range 0 to 31.
Each Eth-Trunk allows a maximum of 64 member
interfaces.
● When 64 is specified, the value is in the range 0 to 63.
Each Eth-Trunk allows a maximum of 32 member
interfaces.
● When 128 is specified, the value is in the range 0 to
127. Each Eth-Trunk allows a maximum of 16 member
interfaces. The default value is 128.
● When 256 is specified, the value is in the range 0 to
255. Each Eth-Trunk allows a maximum of 8 member
interfaces.
● When 512 is specified, the value is in the range 0 to
511. Each Eth-Trunk allows a maximum of 4 member
interfaces.
● When 1024 is specified, the value is in the range 0 to
1023. Each Eth-Trunk allows at most 2 member
interfaces.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 122

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Product Model Value Range

Other models: The value range can be adjusted using the assign forward
eth-trunk mode { 64 | 128 | 256 | 512 | 1024 } command.
● When 64 is specified, the value is in the range 0 to 63.
Each Eth-Trunk allows a maximum of 32 member
interfaces.
● When 128 is specified, the value is in the range 0 to
127. Each Eth-Trunk allows a maximum of 16 member
interfaces. The default value is 128.
● When 256 is specified, the value is in the range 0 to
255. Each Eth-Trunk allows a maximum of 8 member
interfaces.
● When 512 is specified, the value is in the range 0 to
511. Each Eth-Trunk allows a maximum of 4 member
interfaces.
● When 1024 is specified, the value is in the range 0 to
1023. Each Eth-Trunk allows at most 2 member
interfaces.

Step 3 Run commit

The configuration is committed.

----End

3.8.3 Setting the Link Aggregation Mode to LACP

Context
Link aggregation can work in manual load balancing mode, static LACP mode, or
dynamic LACP mode.
In LACP mode, you must manually create an Eth-Trunk and add interfaces to the
Eth-Trunk. However, LACP controls active interfaces through negotiation.
Eth-Trunks in dynamic LACP mode are used only when a Huawei device connects
to a server. For other scenarios, configuring Eth-Trunks in static LACP mode is
recommended to reduce the chances of potential loops.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
Step 3 Configure a working mode of the Eth-Trunk.
Configure the static or dynamic LACP mode.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 123

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● When the two directly connected devices support LACP, static LACP is
recommended.
Run mode lacp-static
The Eth-Trunk is configured to work in static LACP mode.
Before configuring an Eth-Trunk, ensure that both devices use the same
working mode.
● When a device is directly connected to a server, the dynamic LACP mode is
often used. You can also use the static LACP mode.
Run mode lacp-dynamic
The Eth-Trunk is configured to work in dynamic LACP mode.

Step 4 Run commit

The configuration is committed.

----End

3.8.4 Adding Member Interfaces to an Eth-Trunk

Context
You can add member interfaces to an Eth-Trunk in the Eth-Trunk interface view or
member interface view.

When adding an interface to an Eth-Trunk, pay attention to the following points:

● On the CE6881, CE5881, CE6881K, CE6820, CE6863K, CE6881E, and CE6863,

an Eth-Trunk contains a maximum of 128 member interfaces. On the
CE9860EI, an Eth-Trunk contains a maximum of 64 member interfaces. other
models, the number of member interfaces in an Eth-Trunk depends on the
assign forward eth-trunk mode command.
● When member interfaces are added to an Eth-Trunk, they must use the
default interface type and cannot be configured with some services or static
MAC address entries.
● Physical Ethernet interfaces, including split interfaces, can be added to an Eth-
Trunk. However, logical interfaces such as Layer 3 interfaces, VLANIF
interfaces, and Eth-Trunks cannot be added to an Eth-Trunk.
● An Ethernet interface can be added to only one Eth-Trunk. To add the
Ethernet interface to another Eth-Trunk, delete it from the Eth-Trunk first.
● Different Ethernet interfaces can be added to the same Eth-Trunk interface.
To enable interfaces operating at different rates to forward traffic after they
are added to an Eth-Trunk in LACP mode, run the lacp mixed-rate link
enable command.
When Eth-Trunk member interfaces work at different rates, the active
interface is selected as follows: select an Actor based on the system priority
and system ID, and then select the active interface based on the port priority
and port ID of the Actor. To configure a specific interface as the active
interface, run the lacp priority command in the interface view to increase the
interface priority.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 124

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● If an interface of the local device is added to an Eth-Trunk, an interface of the

remote device directly connected to the interface of the local device must also
be added to an Eth-Trunk so that the two ends can communicate.
● After interfaces are added to an Eth-Trunk, the Eth-Trunk learns MAC
addresses and ARP entries but member interfaces do not.
● Devices at both ends of an Eth-Trunk must use the same number of physical
interfaces, interface rate, duplex mode, jumbo and flow control mode.
● In a stack scenario, it is recommended that the number of member interfaces
added to an LAG be the nth power of 2. Otherwise, unknown unicast traffic
may be unevenly load balanced.

Procedure
● Add member interfaces to an Eth-Trunk in the Eth-Trunk interface view.
a. Run system-view
The system view is displayed.
b. Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
c. (Optional) Run lacp mixed-rate link enable
Interfaces operating at different rates are enabled to forward packets
after the interfaces are added to an Eth-Trunk interface in static LACP
mode.

NOTE

The rate of the interface added to the Eth-Trunk interface is not limited. For
example, 10G and 100G interfaces can be added to the same Eth-Trunk interface.
d. Run trunkport interface-type { interface-number1 [ to interface-
number2 ] } &<1-n>
NOTE

The value of n is 128.

A member interface is added to the Eth-Trunk.

When you add member interfaces to an Eth-Trunk in a batch, if one
interface cannot be added to the Eth-Trunk, all subsequent interfaces in
the batch cannot be added to the Eth-Trunk, either.
e. Run commit
The configuration is committed.
● Add member interfaces to an Eth-Trunk in the member interface view.
a. Run system-view
The system view is displayed.
b. Run interface interface-type interface-number
The member interface view is displayed.
c. Run eth-trunk trunk-id
The member interface is added to an Eth-Trunk.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 125

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

d. Run commit
The configuration is committed.
----End

Follow-up Procedure
When the status of an Eth-Trunk member interface changes, the system sends
traps containing the status change information for you to confirm whether the
device encounters any fault. If you want to know the ID of the Eth-Trunk to which
the member interface belongs, run the trunk-member trap in private-mib
enable command to enable Eth-Trunk member interfaces to send traps through a
private MIB. Traps sent through a private MIB carry the number of the specific
Eth-Trunk interface.
After the trunk-member trap in private-mib enable command is run, traps are
sent through a private MIB, instead of a public MIB. In this case, you can view the
traps only by using the private MIB of Huawei.

3.8.5 (Optional) Setting the Upper and Lower Thresholds for

the Number of Active Interfaces
Context
The number of Up member links affects the status and bandwidth of an Eth-
Trunk. To ensure that the Eth-Trunk functions properly and is less affected by
member link status changes, set the following thresholds.
● Lower threshold for the number of active interfaces: When the number of
active interfaces falls below this threshold, the Eth-Trunk goes Down. This
guarantees the Eth-Trunk a minimum available bandwidth.
● Upper threshold for the number of active interfaces: When the number of
active interfaces reaches this threshold, you can add new member interfaces
to the Eth-Trunk, but excess member interfaces enter the Down state. This
improves network reliability with assured bandwidth.
The upper threshold for the number of active interfaces at the local end can be
different from that at the remote end. If the two values are different, the smaller
one is used.
NOTE

After the upper and lower thresholds for the number of active interfaces are set, the range
of the active interface quantity is specified. Depending on the negotiation result of the local
and remote ends on a link, the number of active interfaces can reach the upper threshold
at most. The active interface is selected as follows: select an Actor based on the system
priority and system ID, and then select the active interface based on the port priority and
port ID of the Actor. To configure a specific interface as the active interface, run the lacp
priority command in the interface view to increase the interface priority.

Procedure
Step 1 Run system-view
The system view is displayed.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 126

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Step 2 Run interface eth-trunk trunk-id

The Eth-Trunk interface view is displayed.
Step 3 Run least active-linknumber link-number
The lower threshold for the number of active interfaces is set.
By default, the lower threshold for the number of active interfaces is 1.
The lower threshold for the number of active interfaces on the local device can be
different from that on the remote device. If the two values are different, the larger
one is used.
Step 4 Run lacp max active-linknumber link-number
The upper threshold for the number of active interfaces is set.
By default, CE6881, CE5881, CE6881K, CE6820, CE6863K, CE6881E, and CE6863,
the upper threshold for the number of active interfaces is 128, for CE9860EI, the
upper threshold for the number of active interfaces is 64.
The upper threshold for the number of active interfaces must be greater than or
equal to the lower threshold for the number of active interfaces.
Step 5 Run commit
The configuration is committed.

----End

3.8.6 (Optional) Configuring the Weight of Load Balancing for

a Member Interface

Context
On an Eth-Trunk interface, you can load balance traffic among member interfaces
according to the weights configured for the member interfaces.
The higher the weight of a member interface, the heavier the load over the
member link. Therefore, you can configure a higher weight for a member interface
so that the member link can carry a heavier load.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run interface interface-type interface-number
The Eth-Trunk member interface view is displayed.
Step 3 Run distribute-weight weight-value
The load balancing weight is configured for the Eth-Trunk member interface.
The default weight of an Eth-Trunk member interface is 1.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 127

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

The total weight of member interfaces in an Eth-Trunk cannot exceed the

maximum number of member interfaces allowed.
Step 4 Run commit
The configuration is committed.

----End

3.8.7 (Optional) Configuring a Load Balancing Mode (CE

Switches Excluding the CE6870EI)

Context
An Eth-Trunk supports per-packet load balancing and per-flow load balancing.
● Per-packet load balancing can improve Eth-Trunk bandwidth efficiency to
ensure even load balancing among equal-cost routes, but cannot prevent
packet mis-sequencing. To ensure packet sequencing, confirm that the device
or terminal receiving traffic supports packet reassembly in case of packet mis-
sequencing. Switches support the following per-packet load balancing modes:
– Random mode: The outbound interface of packets is generated randomly
and calculated based on the time when the packets reach the Eth-Trunk.
When the IP address and MAC address of known unicast packets remain
unchanged, configure random per-packet load balancing.
– Round-robin mode: Eth-Trunk member interfaces forward traffic in turn.
When known unicast packets have a similar length, configure round-
robin per-packet load balancing.
● Per-flow load balancing ensures that packets of the same data flow are
forwarded on the same physical link and those of different data flows are
forwarded on different physical links. Table 3-8 lists the load balancing
modes for different types of packets.

Table 3-8 Load balancing modes for different types of packets

Packet Default Load Configurable Remarks

(Inbound Balancing Mode Load Balancing
Interface) Mode

IPv4 packets src-ip, dst-ip, l4- src-ip, dst-ip, l4- The load
src-port and l4- src-port, l4-dst- balancing mode
dst-port port, and is relevant to the
protocol packet type and
irrelevant to the
IPv6 packets src-ip, dst-ip, l4- src-ip, dst-ip, packet
src-port, and l4- protocol, l4-src- forwarding
dst-port port, and l4-dst- process.
port
For example,
even if the
system provides
only Layer 2
forwarding for

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 128

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Packet Default Load Configurable Remarks

(Inbound Balancing Mode Load Balancing
Interface) Mode

MPLS packets Ingress/Egress/ For switches IPv4 packets, the

Transit: top- excluding the IPv4 packets are
label and 2nd- CE6865EI, load balanced
label CE6857EI, according to the
CE6857E, load balancing
CE8861EI, mode for IPv4
CE8868EI: packets. When
Ingress/Egress/ the system
Transit: top- cannot identify
label, 2nd-label, IPv4, IPv6, or
dst-ip, and src-ip MPLS packets,
the system load
For the balances packets
CE6857EI, based on src-
CE6857E, mac, dst-mac,
CE6865EI, src-interface,
CE8861EI, and and eth-type for
CE8868EI: Layer 2 packets.
Ingress/Egress/
Transit: top-
label, 2nd-label,
3rd-label, 4th-
label, 5th-label,
dst-ip, src-ip,
and src-
interface

Layer 2 packets src-mac and dst- src-mac, dst-

except IPv4, IPv6, mac mac, src-
and MPLS interface, and
packets eth-type

TRILL packets Ingress node: src-mac, dst- TRILL packets

inner src-mac mac, src-ip, dst- can be load
and dst-mac for ip, src-interface, balanced on the
Layer 2 packets; l4-src-port, l4- transit node only
src-ip, dst-ip, l4- dst-port when the load-
src-port, and l4- balance
dst-port for enhanced
Layer 3 packets [ resilient ]
profile profile-
name command
is used.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 129

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Packet Default Load Configurable Remarks

(Inbound Balancing Mode Load Balancing
Interface) Mode

Transit/Egress src-mac, dst-

node: inner src- mac, src-ip, dst-
mac and dst- ip, l4-src-port
mac for Layer 2 and l4-dst-port
packets; src-ip,
dst-ip, l4-src-
port, and l4-dst-
port for Layer 3
packets

FCoE packets dst-fcid and src- dst-fcid and src- -

fcid fcid

NOTE

Load balancing is valid only for outgoing traffic; therefore, the load balancing modes for the
interfaces at both ends of the link can be different and do not affect each other.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 (Optional) Run load-balance profile profile-name
A load balancing profile is configured and its view is displayed. profile-name
specifies the name of the load balancing profile.
By default, there is a load balancing profile named default.
Run the following commands as required. You can configure a load balancing
mode for Layer 2 packets, IPv4 packets, IPv6 packets, FCoE, and MPLS packets
respectively.

● Run l2 [ src-mac | dst-mac | src-interface | eth-type ] *

A load balancing mode is configured for Layer 2 packets (non-IP packets).

By default, the switch load balances Layer 2 packets (non-IP packets) based
on the source MAC address (src-mac) and destination MAC address (dst-
mac).
● Run ip [ src-ip | dst-ip | l4-src-port | l4-dst-port | protocol ] *

A load balancing mode is configured for IPv4 packets.

By default, the switch load balances IPv4 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
● Run ipv6 [ src-ip | dst-ip | protocol | l4-src-port | l4-dst-port ] *

A load balancing mode is configured for IPv6 packets.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 130

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

By default, the switch load balances IPv6 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
● Run mpls [ top-label | dst-ip | src-ip | 2nd-label | 3rd-label | 4th-label | 5th-
label | src-interface ] *
A load balancing mode is configured for MPLS packets.
By default, the switch load balances MPLS packets based on the two outer
labels (top-label and 2nd-label).
NOTE

Only the CE6857EI, CE6857E, CE6865EI, CE8861EI, and CE8868EI support 3rd-label,
4th-label, 5th-label, and src-interface. If src-ip or dst-ip is configured, 4th-label or
5th-label cannot be configured.
In V200R005C10 and earlier versions, if 4th-label and 5th-label are configured for
load balancing of MPLS packets in the load balancing profile view or ECMP view, this
configuration takes effect for both Eth-Trunk and ECMP. The load balancing mode
based on src-ip and dst-ip conflicts with that based on 4th-label and 5th-label, and
the last delivered configuration takes effect. Therefore, the effective load balancing
mode is inconsistent with that in the configuration file.
● Run fcoe { dst-fcid | src-fcid } *

A load balancing mode is configured for FCoE packets.

By default, the switch load balances FCoE packets based on the source FC_ID
(src-fcid) and destination FC_ID (dst-fcid).
● Run tunnel {outer-header | inner-header}
Load balancing based on the inner or outer header is configured for tunnel
packets (GRE and IP-in-IP packets) in the specified load balancing profile.
By default, GRE and IP-in-IP packets are load balanced based on inner
headers.
NOTE

This command is supported on the following models: CE5855E, CE6856HI, CE6857EI,

CE6857E, CE6865EI, CE8850EI, CE8861EI, CE8868EI, CE9860EI.
● Run eth-trunk { hash-mode hash-mode-id | universal-id universal-id } *

A load balancing mode is configured for the Eth-Trunk.

By default, an Eth-Trunk load balances packets based on hash-mode (1) and
universal-id (1).
● Run stack { hash-mode hash-mode-id | universal-id universal-id } *

A load balancing mode is configured for a stack port or fabric port in the
specified load balancing profile.
By default, a stack port or fabric port load balances packets based on hash-
mode (1) and universal-id (1) in a load balancing profile.

Step 3 (Optional) Run quit

Return to the system view.

Step 4 Run the following commands as required.

● Configure a load balancing mode for known unicast traffic.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 131

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

a. Run interface eth-trunk trunk-id

The Eth-Trunk interface view is displayed.
b. Run load-balance { dst-ip | dst-mac | random | round-robin | src-ip |
src-mac | src-dst-ip | src-dst-mac | enhanced [ resilient ] profile profile-
name }
A load balancing mode is configured for the Eth-Trunk.
By default, the load balancing mode of the Eth-Trunk is enhanced
profile.
NOTE

Only the CE8868EI, CE8861EI, CE6881K, CE6881E, CE6881, CE5881, CE6865EI,

CE6863K, CE6863, CE6857EI, CE6857E, CE6856HI , CE6820 support resilient
parameter.
Only the CE8868EI, CE8861EI, CE8850EI, CE6865EI, and CE6857EI, CE6857E support
random and round-robin parameters.
CE9860EI only supports random and enhanced parameters.

▪ dst-ip: based on destination IP addresses

▪ dst-mac: based on destination MAC addresses

▪ random: The outbound interface of packets is generated randomly

and calculated based on the time when the packets reach the Eth-
Trunk. When the IP address and MAC address of known unicast
packets do not change, use this mode to perform packet-based load
balancing so that traffic is load balanced. The random mode causes
the mis-sequencing problem. Ensure that the receive device or
terminal supports assembly of mis-sequenced packets.

▪ round-robin: Each Eth-Trunk member interface forwards traffic in

turn. For known unicast traffic, if packets have approximate lengths,
configure this load balancing mode to achieve even load balancing.
This mode may cause the mis-sequencing problem. Ensure that the
receive device or terminal supports assembly of mis-sequenced
packets.

▪ src-ip: based on source IP addresses

▪ src-mac: based on source MAC addresses

▪ src-dst-ip: based on the Exclusive-Or result of source and destination

IP addresses

▪ src-dst-mac: based on the Exclusive-Or result of source and

destination MAC addresses

▪ enhanced [ resilient ] profile: based on the fields in the global load

balancing profile. resilient indicates that traffic on links is switched
less as much as possible when links are increased or reduced and
only some traffic is switched between links. For example, an LAG has
three member links, data is forwarded based on the hash key, and
one link is faulty. When resilient is not specified, traffic is reallocated
on the other two links. When resilient is specified, traffic that has
been allocated on the other two links remain unchanged, and traffic

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 132

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

on the faulty link is evenly allocated on the two links. Therefore,

there is less impact on services. When the faulty link recovers, some
traffic on the other two links is switched to it. Traffic allocation on
links is different from that before the fault occurrence.
c. Run commit
The configuration is committed.
● Configure a load balancing mode for unknown unicast traffic.
a. Run load-balance unknown-unicast { mac | enhanced }
A load balancing mode is configured.
By default, the load balancing mode is enhanced.
b. Run commit
The configuration is committed.

----End

Follow-up Procedure
● Configure simulated calculation of an Eth-Trunk outbound interface.
Configure simulated calculation of an Eth-Trunk outbound interface after
specifying the 5-tuple information, source MAC address, and destination
address.
display load-balance forwarding-path unicast interface eth-trunk trunk-id src-interface interface-
type interface-number { ethtype ethtype-number | vlan vlan-id | [ [ src-ip src-ip-data | dst-ip dst-ip-
data ] * | [ src-ipv6 src-ipv6-data | dst-ipv6 dst-ipv6-data ] * ] | src-mac src-mac-data | dst-mac dst-
mac-data | protocol { protocol-number | icmp | igmp | ip | ospf | tcp [ l4-src-port src-port-data | l4-
dst-port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-dst-port dst-port-data ] * } } * slot slot-
id

● Verifying the Configuration

Verify the outbound interface of packets that contain specified 5-tuple
information, source MAC address, and destination MAC address.
display port forwarding-path { src-ip src-ip-data | dst-ip dst-ip-data | src-mac src-mac-data | dst-
mac dst-mac-data | protocol { protocol-number | gre | icmp | igmp | ip | ipinip | ospf | tcp [ l4-src-
port src-port-data | l4-dst-port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-dst-port dst-
port-data ] * } } *

3.8.8 (Optional) Configuring a Load Balancing Mode

(CE6870EI)

Context
An Eth-Trunk uses flow-based load balancing. Per-flow load balancing ensures
that packets of the same data flow are forwarded on the same physical link and
those of different data flows are forwarded on different physical links.

Load balancing is valid only for outgoing traffic; therefore, interfaces at both ends
of the link can use different load balancing modes.

Table 3-9 lists load balancing modes for different types of packets.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 133

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Table 3-9 Load balancing modes for different types of packets

Packet (Inbound Default Load Balancing Configurable Load
Interface) Mode Balancing Mode

IPv4 packets src-ip, dst-ip, l4-src- src-ip, dst-ip, l4-src-

port, and l4-dst-port port, l4-dst-port, and
protocol

IPv6 packets src-ip, dst-ip, l4-src- src-ip, dst-ip, l4-src-

port, and l4-dst-port port, l4-dst-port, flow-
label, and protocol
NOTE
IPv6 packet load balancing
modes, l4-src-port and l4-
dst-port, are affected by
the l4-src-port and l4-dst-
port fields of IPv4 packets.
That is, when the load
balancing modes of IPv4
packets include l4-src-port
or l4-dst-port, the l4-src-
port or l4-dst-port field
also participates in load
balancing of IPv6 packets.

MPLS packets Ingress/Egress/Transit: Ingress/Egress/Transit:

top-label, 2nd-label, top-label, 2nd-label,
fields in the inner IP and 3rd-label
header (IPv4: src-ip, dst- NOTE
ip, l4-src-port, and l4- The load balancing mode
dst-port; IPv6: src-ip, based on fields in the inner
dst-ip, l4-src-port, and IP header (IPv4: src-ip, dst-
ip, l4-src-port, and l4-dst-
l4-dst-port), and 2nd-
port; IPv6: src-ip, dst-ip,
label l4-src-port, and l4-dst-
port) does not need to be
configured, but the fields
that participate in load
balancing are affected by
the load balancing mode
configuration of IPv4 and
IPv6 packets. If the l4-src-
port and l4-dst-port fields
are specified in the
configured load balancing
mode, packets may be
unable to be load balanced
in the configured load
balancing mode. To
prevent this problem,
cancel the configuration of
the l4-src-port and l4-dst-
port fields in the load
balancing mode of IPv4
and MPLS packets.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 134

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Packet (Inbound Default Load Balancing Configurable Load

Interface) Mode Balancing Mode

VPLS packets Ingress/Egress/Transit: Transit node: inner-sip

top-label, fields in the and inner-dip
inner ETH header (src- PE node: inner-ip and
mac, dst-mac, and ETH inner-ipv6
TYPE), and 2nd-label
NOTE
For VPLS packets, the load
balancing mode based on
fields in the inner Ethernet
header (src-mac, dst-mac
and eth-type) does not
need to be configured, but
the fields that participate
in load balancing are
affected by the load
balancing mode
configuration of the l2
[ src-mac | dst-mac | vlan
| eth-type ] * command.

Layer 2 packets except src-mac, dst-mac and src-mac, dst-mac, vlan,

IPv4, IPv6, and MPLS vlan and eth-type
packets

TRILL packets Ingress node: src-mac, src-mac, dst-mac, src-ip,

dst-mac, and vlan for dst-ip, l4-src-port, l4-
Layer 2 packets; src-ip, dst-port, and protocol
dst-ip, l4-src-port, and
l4-dst-port for Layer 3
packets

Transit node: src-mac, src-mac, dst-mac, vlan,

dst-mac, vlan, src-ip, src-ip, and dst-ip
and dst-ip in the inner
tag

Egress node: src-mac, src-mac, dst-mac, vlan,

dst-mac, vlan, src-ip, src-ip, and dst-ip
and dst-ip in the inner
tag

FCoE packets dst-fcid and src-fcid dst-fcid, src-fcid

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run load-balance profile profile-name
A load balancing profile is configured and its view is displayed. profile-name
specifies the name of the load balancing profile.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 135

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

By default, there is a load balancing profile named default.

Step 3 Run the following commands as required. You can configure load balancing modes
for Layer 2, IPv4, IPv6, MPLS, VPLS, TRILL, and FCoE packets, respectively.
● Run l2 [ src-mac | dst-mac | vlan | eth-type ] *

A load balancing mode is configured for Layer 2 packets (not IP packets) in

the specified load balancing profile.
NOTE

During Layer 2 forwarding, the switch load balances only IPv4, IPv6, 802.1ah, ARP, and
CFM packets based on eth-type.
During Layer 2 forwarding, an Eth-Trunk configured with Dot1q tunnel, VLAN stacking,
VLL, or VPLS cannot load balance packets based on the VLAN ID.
● Run ip [ src-ip | dst-ip | l4-src-port | l4-dst-port | protocol ] *

A load balancing mode is configured for IPv4 packets.

By default, the switch load balances IPv4 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
● Run ipv6 [ src-ip | dst-ip | protocol | flow-label ] *

A load balancing mode is configured for IPv6 packets.

By default, the switch load balances IPv6 packets based on the source IP
address (src-ip), destination IP address (dst-ip), transport-layer source port
numbers (l4-src-port), and transport-layer destination port numbers (l4-dst-
port).
NOTE

To load balance IPv6 packets based on the transport-layer source and destination port
numbers, run the ip [ l4-src-port | l4-dst-port ] * command.
● Run mpls [ 2nd-label | 3rd-label | top-label ] * or mpls [ src-ip | dst-ip | l4-
src-port | l4-dst-port ] *
A load balancing mode is configured for MPLS packets in the specified load
balancing profile.
By default, MPLS packets are load balanced based on top-label, 2nd-label,
and fields in the inner IP header (IPv4/IPv6: src-ip, dst-ip, l4-src-port, and l4-
dst-port).

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 136

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

On an IPv6 underlay network, if the IPv6 VXLAN function is enabled and the switch
functions as a transit or egress node on the MPLS network, configurations related to
the hash field in inner information in MPLS packets do not take effect in the load
balancing profile. In this scenario, the mpls inner-type { ipv4 | ipv6 | l2 | any }
command cannot be used to change the load balancing mode for MPLS packets in the
load balancing profile. Instead, you can use the source and destination IP addresses in
inner information in MPLS packets to configure a load balancing mode in the load
balancing profile.
– The mpls [ inner-ip | inner-ipv6 ] * command can be configured only when the
IPv6 VXLAN function is enabled. In addition, this command cannot be used to
configure a load balancing mode for VPLS packets. If the IPv6 VXLAN function is
disabled, the switch automatically deletes the configurations related to the mpls
[ inner-ip | inner-ipv6 ] * command.
– For the CE6870EI, if MPLS packets with four or more labels are received, Eth-Trunk
or ECMP load balancing cannot be implemented based on the inner IP header.
– For the CE6870EI, if the implicit null label is disabled and the outer label is popped
out (for example, ASBR in inter-AS VPN Option B and inter-AS VPN Option C
networking), packets may be unable to be load balanced in the load balancing
mode using ECMP or Eth-Trunk based on the inner field in MPLS packets.
● Run user-defined ethernet-over-mpls [ dot1q-tagtag-number ] { inner-
sipsip-offsetsip-nybble-number | inner-dipdip-offsetdip-nybble-number } *
Load balancing based on the inner IP address of VPLS packets is configured
on the outbound Eth-Trunk on the transit node.
● Run user-defined ethernet-over-mpls { inner-ip | inner-ipv6 }
Load balancing based on the inner IP address of VPLS packets is configured
on the outbound Eth-Trunk on the PE node.
NOTE

If inner-ip is specified, you need to run ip [ src-ip | dst-ip ] *. If inner-ipv6 is specified,

you need to run ipv6 [ src-ip | dst-ip ] *. Otherwise, the user-defined ethernet-over-
mpls { inner-ip | inner-ipv6 } command does not take effect.
● Run fcoe { dst-fcid | src-fcid } *

A load balancing mode is configured for FCoE packets.

By default, the switch load balances FCoE packets based on the source FC_ID
(src-fcid) and destination FC_ID (dst-fcid).
● Run trill egress mode { l2 | ipv4 | ipv6 }
An Eth-Trunk load balancing mode is configured on egress and transit nodes
of the TRILL network.
By default, the Eth-Trunk load balancing mode on egress and transit nodes of
the TRILL network is ipv4.
– l2: indicates load balancing based on the source MAC address,
destination MAC address, and VLAN ID. The items used in load balancing
depend on the l2 command.
– ipv4: indicates load balancing based on the source MAC address,
destination MAC address, source IPv4 address, destination IPv4 address,
and VLAN ID. The items used in load balancing depend on the l2 and ip
commands.
– ipv6: indicates load balancing based on the source MAC address,
destination MAC address, source IPv6 address, destination IPv6 address,

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 137

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

and VLAN ID. The items used in load balancing depend on the l2 and
ipv6 commands.
● Run tunnel {outer-header | inner-header}*
Load balancing based on the inner or outer header is configured for tunnel
packets (GRE and IP-in-IP packets) in the specified load balancing profile.
By default, GRE and IP-in-IP packets are load balanced based on both inner
and outer headers.
NOTE

Only CE6870EI Supports this command.

● Run ip-tunnel { inner-src-ip | inner-dst-ip | inner-l4-src-port | inner-l4-dst-
port | inner-protocol } *
Load balancing based on the inner header is configured for tunnel packets
(GRE and IP-in-IP packets) in the specified load balancing profile.
NOTE

– This command does not take effect when GRE and IP-in-IP packets are load
balanced based on the outer header.
– Only CE6870EI Supports this command.
● Run eth-trunk { src-interface | seedseed-data | universal-iduniversal-id |
hash-modehash-mode-id } *
An Eth-Trunk load balancing mode is configured in the specified load
balancing profile.
NOTE

Packets forwarded at Layer 3 cannot be load balanced based on destination MAC

addresses, source MAC addresses, VLAN IDs, and Ethernet types.
Eth-Trunk and ECMP on the device share the same load balancing profile. The load
balancing mode configured using the l2, ip, ipv6, or mpls command takes effect for
both link aggregation and ECMP. The load balancing modes configured using the eth-
trunk (load-balance-profile view) and ecmp (load-balance-profile view)
commands take effect only for link aggregation and ECMP respectively.
When load balancing using both Eth-Trunk and ECMP is uneven, change the value of
universal-id or the hash mode.
● Run mode symmetry
A method is configured for an Eth-Trunk to hash a pair of flows to the same
interface.
By default, no method is configured for an Eth-Trunk to hash a pair of flows
to the same interface.
Step 4 Run commit
The configuration is committed.

----End

Follow-up Procedure
● Configure simulated calculation of an Eth-Trunk outbound interface.
Run the display load-balance forwarding-path unicast interface eth-trunk
trunk-id src-interface interface-type interface-number { ethtype ethtype-

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 138

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

number | vlan vlan-id | [ [ src-ip src-ip-data | dst-ip dst-ip-data ] * | [ src-ipv6

src-ipv6-data | dst-ipv6 dst-ipv6-data | flow-label label-data ] * ] | src-mac
src-mac-data | dst-mac dst-mac-data | protocol { protocol-number | icmp |
igmp | ip | ospf | tcp [ l4-src-port src-port-data | l4-dst-port dst-port-data ] *
| udp [ l4-src-port src-port-data | l4-dst-port dst-port-data ] * } } * slot slot-id
command to configure simulated calculation of an Eth-Trunk outbound
interface after the 5-tuple information, source MAC address, and destination
address are specified.
● Verify the configuration.
Run the display port forwarding-path { src-ip src-ip-data | dst-ip dst-ip-data
| src-mac src-mac-data | dst-mac dst-mac-data | protocol { protocol-number |
gre | icmp | igmp | ip | ipinip | ospf | tcp [ l4-src-port src-port-data | l4-dst-
port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-dst-port dst-port-
data ] * } } * [ enhanced ] command to check the outbound interface of
packets that contain specified 5-tuple information, source MAC address, and
destination MAC address.
Run the display port forwarding-path mpls { { src-ip src-ip-data [ ip-mask-
len | source-ip-mask ] | dst-ip dst-ip-data [ ip-mask-len | dst-ip-mask ] } * |
{ src-ipv6 src-ipv6-data [ ipv6-mask-len ] | dst-ipv6 dst-ipv6-data [ ipv6-
mask-len ] } * }{ transit label-number labelnum | ingress | egress label-
number labelnum } command to check the outbound interface of MPLS
packets that contain information such as the inner source IP address,
destination IP address, role, and number of labels.

3.8.9 (Optional) Configuring an Eth-Trunk Load Balancing

Mode for PPPoE Packets

Context
By default, a CE switch processes received PPPoE packets as common Ethernet
packets. Figure 3-16 shows the format of PPPoE packets. The switch cannot
identify the 5-tuple information in PPPoE packets and performs the hash
algorithm based on the outer Ethernet frame of common Layer 2 packets by
default. Fields in the Layer 2 frames of PPPoE packets are fixed except the source
MAC address, so packets are often unevenly load balanced on an Eth-Trunk. To
improve the load balancing effect, use inner information of PPPoE packets for load
balancing.

Figure 3-16 Format of PPPoE packets

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 139

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

The CE6881, CE5881, CE6881K, CE6820, CE6863, CE6863K, CE6881E can identify PPPoE
packets and load balance the PPPoE packets without configuring Eth-Trunk load balancing.

Procedure
● Configure a load balancing mode for PPPoE packets on the CE6870EI.
a. Run system-view
The system view is displayed.
b. Run load-balance eth-trunk pppoe { session-id | l4-src-port { ppp-
address-compression | ppp-protocol-compression | both | none } }
A load balancing mode is configured for PPPoE packets.
By default, the switch load balances PPPoE packets based on the source
MAC address (smac), destination MAC address (dmac), and VLAN ID
(vlan).
You can specify session-id and l4-src-port so that the switch load
balances PPPoE packets based on the session ID and transport-layer
source port of PPPoE packets.
c. Run commit
The configuration is committed.
● Configure a load balancing mode for PPPoE packets on other models
excluding the CE6870EI.
a. Run system-view
The system view is displayed.
b. Run load-balance profile profile-name
The global load balancing profile view is displayed.
By default, the load balancing profile name is default.
c. Run eth-trunk hash-mode 8
The Eth-Trunk load balancing hash algorithm is set to 8.
d. Run quit
The system view is displayed.
e. Run load-balance ecmp
The ECMP view is displayed.
f. Run hashmode 7
The ECMP load balancing hash algorithm is set to 7.
g. Run quit
The system view is displayed.
h. Run load-balance pppoe { session-id | l4-src-port { ppp-address-
compression | ppp-protocol-compression | both | none } }
A load balancing mode is configured for PPPoE packets.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 140

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

By default, the switch load balances PPPoE packets based on the source
MAC address (smac) and destination MAC address (dmac).
You can specify session-id and l4-src-port so that the switch load
balances PPPoE packets based on the session ID and transport-layer
source port of PPPoE packets.

NOTE

This command is supported only by the following models: CE6856HI, CE6857EI,

CE6857E, CE6865EI, CE9860EI, and CE8800 series switches.
i. Run commit
The configuration is committed.
----End

3.8.10 (Optional) Setting the LACP System Priority

Context
LACP system priorities determine the sequence in which devices at two ends of an
Eth-Trunk select active interfaces to join a LAG. In order for a LAG to be
established, both devices must select the same interfaces as active interfaces. To
achieve this, one device (with a higher priority) is responsible for selecting the
active interfaces. The other device (with a lower priority) then selects the same
interfaces as active interfaces.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run lacp priority priority
The LACP system priority is set.
In priority comparisons, numerically lower values have higher priority. By default,
the LACP system priority is 32768.
The end with a smaller priority value functions as the Actor. If the two ends have
the same priority, the end with a smaller MAC address functions as the Actor.
Step 3 Run commit
The configuration is committed.

----End

3.8.11 (Optional) Setting the LACP Interface Priority

Context
In LACP mode, LACP interface priorities are set to prioritize interfaces of the same
device. Interfaces with higher priorities are selected as active interfaces.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 141

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run interface interface-type interface-number

The member interface view is displayed.

Step 3 Run lacp priority priority

The LACP priority of the member interface is configured.

By default, the LACP interface priority is 32768. In priority comparisons,

numerically lower values have higher priority.

By default, the system selects active interfaces based on interface priorities. This
may lead to low-speed member interfaces with high priorities being selected as
active interfaces, providing less bandwidth for services. To select high-speed
member interfaces as active interfaces, run the lacp select { priority | speed }
command to configure the system to select active interfaces based on the
interface rate.

Step 4 Run commit

The configuration is committed.

----End

3.8.12 (Optional) Setting the LACP System ID and MAC

Address in Layer 3 Mode

Context
When devices are connected through Eth-Trunk interfaces in LACP mode, the
device with a higher LACP system priority functions as the LACP Actor. The other
device then selects active member interfaces based on the interface priorities of
the LACP Actor. If the two connected devices have the same LACP system priority,
the LACP system IDs determine the device priorities. To configure an LACP system
ID, run the lacp system-id command. The device with a higher priority then
becomes the LACP Actor. The other device then selects active member interfaces
based on the interface priorities of the LACP Actor.

In this scenario, two Layer 3 Eth-Trunks of the device use the bridge MAC address
by default. When the MAC address is the same, the Layer 3 protocol cannot work.
To ensure that Layer 3 traffic is forwarded normally, configure the MAC address in
Layer 3 mode for at least one Eth-Trunk so that the two Eth-Trunks use different
MAC addresses.

Procedure
Step 1 Run system-view

The system view is displayed.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 142

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Step 2 Run interface interface-type interface-number

The member interface view is displayed.
Step 3 Run lacp system-id mac-address
The LACP system ID of the member interface is configured.
By default, the LACP system ID of an Eth-Trunk interface is the system bridge MAC
address.
The lacp system-id and lacp m-lag system-id commands are mutually exclusive.
Step 4 (This step is mandatory when Layer 3 traffic needs to be forwarded.) Configure
the MAC Address for an Eth-Trunk in Layer 3 Mode.
1. Run undo portswitch
The Eth-Trunk changes to the Layer 3 mode.
By default, an Eth-Trunk works in Layer 2 mode.
2. Run mac-address mac-address
The MAC address of the Eth-Trunk is configured.
By default, an Eth-Trunk uses the system MAC address.
Step 5 Run commit
The configuration is committed.

----End

3.8.13 (Optional) Configuring LACP Preemption

Context
The LACP preemption function ensures that the interface with the highest LACP
priority will return to functioning as an active interface after recovering from a
fault. If LACP preemption is disabled, the interface cannot become active interface
after it recovers.
The LACP preemption delay is the period that an inactive interface waits before
changing to the active interface. The LACP preemption delay reduces the chances
of unstable data transmission on an Eth-Trunk due to frequent status changes of
some links.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 (Optional) Run lacp ignore aggregation delay
The device is enabled to ignore the value of the Reserved field in received
LACPDUs.
By default, the device identifies the value of the Reserved field in received
LACPDUs.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 143

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

When a Huawei device is connected to a non-Huawei device, if the Reserved field

in LACPDUs defined by other vendors is different from that defined by Huawei,
LACP flapping occurs and services are interrupted. In this scenario, you are advised
to configure the device to ignore the value of the Reserved field in received
LACPDUs and set the delay on both ends to the same value.
Step 3 Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
Step 4 Run lacp preempt enable
LACP preemption is enabled for the Eth-Trunk.
By default, LACP preemption is disabled. To ensure normal running of an Eth-
Trunk, enable or disable LACP preemption on both ends of the Eth-Trunk.
Step 5 Run lacp preempt delay delay-time
The LACP preemption delay of the Eth-Trunk is set.
By default, the LACP preemption delay is 30 seconds. If both devices of an Eth-
Trunk use different preemption delays, the longer preemption delay becomes
effective.
Step 6 Run commit
The configuration is committed.

----End

3.8.14 (Optional) Setting the Timeout Interval for Receiving

LACPDUs
Context
If the Eth-Trunk on the local device cannot detect a self-loop or fault that occurred
on a member interface in the LAG on the remote device, data on the local device
is still load balanced among original active interfaces. As a result, data traffic on
the faulty link is discarded.
After the timeout interval at which LACPDUs are received is set, if a local member
interface does not receive any LACPDUs within the configured timeout interval,
the local member interface becomes Down immediately and no longer forwards
data.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run interface eth-trunk trunk-id
The Eth-Trunk interface view is displayed.
Step 3 Run lacp timeout { fast [ user-defined user-defined ] | slow }
The timeout interval at which LACPDUs are received is set.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 144

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

By default, the timeout interval at which an Eth-Trunk receives LACPDUs is 90

seconds.
● After you run the lacp timeout command, the local end notifies the remote
end of the timeout interval by sending LACPDUs. When fast is specified, the
interval for sending LACPDUs is 1 second. When slow is specified, the interval
for sending LACPDUs is 30 seconds.
● When fast is specified, the timeout interval for receiving LACPDUs is 3
seconds. When slow is specified, the timeout interval for receiving LACPDUs is
90 seconds.
● You can use different modes of the timeout interval at the two ends.
However, to facilitate maintenance, you are advised to use the same mode at
both ends.
Step 4 Run commit
The configuration is committed.
----End

3.8.15 (Optional) Binding a VLAN to an Eth-Trunk Member

Interface

Context
A server often has two or more NICs. One NIC is used for server management and
service traffic forwarding, other NICs are used for only service traffic forwarding. In
Figure 3-17, a switch is connected to a server through the Eth-Trunk that is added
to VLAN 10 and VLAN 20. VLAN 10 and VLAN 20 are used for forwarding
management packets and service packets, respectively. An independent VLAN
needs to be configured on an Eth-Trunk member interface that connected to the
management NIC of a server, that is, VLAN 10 needs to be bound to the Eth-Trunk
member interface so that the Eth-Trunk member interface can be specially used
for server management. This prevents management packets from being discarded
when management packets are load balanced on different NICs. In addition,
service traffic of VLAN 20 can still be load balanced on all Eth-Trunk member
interfaces, improving the bandwidth usage.

Figure 3-17 Connecting a switch to a server

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 145

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

When the Eth-Trunk member interface configured with the independent VLAN is
in Unselect state during LACP negotiation, it cannot forward packets. In this
situation, you can run the trunk member vlan lacp disable command to
configure a flexibly independent VLAN on an Eth-Trunk member interface. When
the Eth-Trunk member interface is in Unselect state, the flexibly independent
VLAN becomes available. The system automatically disables LACP so that the Eth-
Trunk member interface in Unselect state can still forward packets. When the
flexibly independent VLAN becomes available, the Eth-Trunk member interface can
only forward packets in the local VLAN but cannot process traffic from other
VLANs. When the Eth-Trunk member interface enters the Select state, the
independent VLAN becomes available and the Eth-Trunk member interface can
still process traffic from other VLANs.

NOTE

The CE9860EI, CE6870EI do not support this command.

NOTE

● This function applies to the scenario where a switch connects to a server, and it can only be
configured on the member interface connecting to the server's management NIC.
● In an Eth-Trunk, only one member interface can be bound to a VLAN or VLANs and one
member interface can be bound to a maximum of eight VLANs. Member interfaces of
multiple Eth-Trunks can be bound to the same VLAN.
● The switch supports a maximum of 256 bound VLANs. If N Eth-Trunk member interfaces
are bound to M VLANs, the maximum value of N multiplied by M is 256.
● After this command is configured, packets from the bound VLAN can be only forwarded
through the bound Eth-Trunk member interface. Non-unicast traffic on the Eth-Trunk can
be only forwarded through the bound Eth-Trunk member interface.
● When an Eth-Trunk member interface is bound to a VLAN or VLANs, VLAN mapping, VLAN
stacking, or MUX VLAN cannot be configured.
● When an Eth-Trunk member interface is bound to a VLAN or VLANs, M-LAG cannot be
configured.

Procedure
● Bind an independent VLAN to an Eth-Trunk member interface.
a. Run system-view

The system view is displayed.

b. Run interface interface-type interface-number

The Eth-Trunk member interface view is displayed.

c. Run trunk member binding vlan vlan-id

A VLAN is bound to the Eth-Trunk member interface.

By default, no VLAN is bound to an Eth-Trunk member interface.

When binding a VLAN to an Eth-Trunk member interface, ensure that the

Eth-Trunk interface has been added to the VLAN.
d. Run commit

The configuration is committed.

● Bind a flexibly independent VLAN to an Eth-Trunk member interface.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 146

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

a. Run system-view

The system view is displayed.

b. Run interface eth-trunk trunk-id

The Eth-Trunk interface view is displayed.

c. Run mode { lacp-static | lacp-dynamic }

The Eth-Trunk is configured to work in LACP mode.

d. Run quit

Exit from the Eth-Trunk interface view.

e. Run interface interface-type interface-number

The Eth-Trunk member interface view is displayed.

f. Run trunk member binding vlan vlan-id

A VLAN is bound to the Eth-Trunk member interface.

By default, no VLAN is bound to an Eth-Trunk member interface.

When binding a VLAN to an Eth-Trunk member interface, ensure that the

Eth-Trunk interface has been added to the VLAN.
g. Run trunk member lacp disable

A flexibly independent VLAN is bound to the Eth-Trunk member interface.

By default, an Eth-Trunk member interface is not configured with a

flexibly independent VLAN.

After this command is configured, all VLANs bound to the Eth-Trunk

member interface become flexibly independent VLANs and the Eth-Trunk
member interface can only forward packets from the flexibly independent
VLANs.
h. Run commit

The configuration is committed.

----End

3.8.16 (Optional) Configuring an Eth-Trunk Member Interface

in Force Up State

Context
When a server connects to the switch, to improve the reliability, the switch
interface directly connected to the server is added to the Eth-Trunk in static LACP
mode. When the server restarts or goes online and the timeout interval of the Eth-
Trunk that receives LACPDUs is reached, the Eth-Trunk member interface becomes
Down. You can configure the Eth-Trunk member interface in Force Up state so that
the Eth-Trunk member interface can continue to forward service traffic.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 147

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run interface interface-type interface-number
The interface view is displayed.
Step 3 Run lacp force-up [ extension ]
The interface is configured in Force Up state.
By default, an interface is not in Force Up state.
After this lacp force-up command is run, the force-up state takes effect only
when all the member interfaces of the Eth-Trunk interface in static LACP mode
time out in receipt of LACPDUs.
When all the Eth-Trunk member interfaces' force-up state takes effect, the
minimum number of active member links configured using the least active-
linknumber link-number command still takes effect, but the maximum number of
active member links configured using the max active-linknumber link-number
command stops taking effect.
To prevent the force-up function from causing packet loss when server and device
interfaces are not correctly connected or the interface is falsely removed from the
Eth-Trunk interface, run the lacp force-up extension command to enable the
extended force-up function on the interface. After the extended force-up function
is enabled, the force-up function can take effect only once. If the interface state
changes from down to up later, the force-up function can take effect once again.
Step 4 Run commit
The configuration is committed.

----End

3.8.17 (Optional) Enabling State Flapping Suppression on an

Eth-Trunk
Context

● If the state of an Eth-Trunk frequently flaps, the LACP protocol status of the
Eth-Trunk also flaps, affecting the Eth-Trunk operations. To resolve this
problem, enable the state flapping suppression function on the Eth-Trunk
working in LACP mode.
● After LACP negotiation succeeds on an Eth-Trunk interface, the interface will
save the source MAC address of the most recently received packet, and will
check the source MAC addresses of packets. If the Eth-Trunk interface receives
a packet with the source MAC address being different from the one which
saved, the Eth-Trunk may be flapped because the once more negotiation.
To prevent an Eth-Trunk interface from alternating between Up and Down,
enable the invalid-MAC-based flapping suppression function on the Eth-Trunk

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 148

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

interface working in LACP mode. After this function is enabled, the Eth-Trunk
interface drops the packet and records the packet information if receiving a
packet with the source MAC address being different from the valid one. In this
manner, the Eth-Trunk interface does not alternate between Up and Down
even if packet flapping occurs.

Procedure
● Enable the state flapping suppression function on an Eth-Trunk.
a. Run system-view

The system view is displayed.

b. Run interface eth-trunk trunk-id

The Eth-Trunk interface view is displayed.

c. Run lacp dampening state-flapping

The state flapping suppression function is enabled on the Eth-Trunk.

By default, the state flapping suppression function is disabled on an Eth-

Trunk.
d. Run commit

The configuration is committed.

● Enable the invalid-MAC-based state flapping suppression function on an Eth-
Trunk.
a. Run system-view

The system view is displayed.

b. Run interface eth-trunk trunk-id

The Eth-Trunk interface view is displayed.

c. Run undo lacp dampening unexpected-mac disable

The invalid-MAC-based state flapping suppression function is enabled on

the Eth-Trunk.

By default, the invalid-MAC-based state flapping suppression function is

enabled on an Eth-Trunk.
d. Run commit

The configuration is committed.

----End

3.8.18 Verifying the Link Aggregation Configuration

Procedure
● Run the display eth-trunk [ trunk-id [ interface interface-type interface-
number | verbose ] | brief ] command to check the Eth-Trunk configuration.
● Run the display eth-trunk membership trunk-id command to check
information about Eth-Trunk member interfaces.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 149

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Run the display load-balance profile [ profile-name ] command to check the

load balancing profile of the Eth-Trunk.
● Run the display forward eth-trunk mode command to check the number of
LAGs supported by the device.
NOTE

The CE9860EI, CE6881, CE5881, CE6881K, CE6820, CE6863, CE6863K, and CE6881E do
not support this command.

----End

3.9 Configuring Preferential Forwarding of Local Traffic

on an Eth-Trunk in a Stack
On a network where interfaces of multiple switches in a stack form an Eth-Trunk,
you can configure the Eth-Trunk to preferentially forward local traffic to increase
bandwidth use efficiency between stack devices and improve traffic forwarding
efficiency.

Context
Determine whether to enable preferentially forwarding of local traffic on an Eth-
Trunk based on actual needs:

● If active interfaces in the local Eth-Trunk have sufficient bandwidth to forward

traffic on the local device, enable this function to improve traffic forwarding
efficiency and increase bandwidth use efficiency between stack devices.
● If active interfaces in the local Eth-Trunk do not have sufficient bandwidth to
forward traffic on the local device, disable this function to prevent packet loss.
Some traffic on the local device is forwarded through member interfaces on
another device.
NOTE

The CE9860EI does not support this function.

Pre-configuration Tasks
Before enabling local preferential forwarding on an Eth-Trunk, complete the
following tasks:

● Create an Eth-Trunk and add physical interfaces to the Eth-Trunk.

● Establish a stack.
● Ensure that member interfaces of the local Eth-Trunk have sufficient
bandwidth to forward local traffic.

Procedure
Step 1 Run system-view

The system view is displayed.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 150

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Step 2 Run interface eth-trunk trunk-id

The view of an Eth-Trunk is displayed.

Step 3 Run undo local-preference disable

The Eth-Trunk is configured to preferentially forward local traffic.

By default, an Eth-Trunk forwards traffic preferentially through local member

interfaces.

NOTE

This function is valid only for known unicast packets, and does not work with unknown
unicast packets, broadcast packets, and multicast packets.

Step 4 (Optional) Run local-preference enhanced

The chip is configured to preferentially forward local traffic.

By default, a chip on a device does not preferentially forward local traffic.

NOTE

Only the CE6863, CE6863K, CE6881E, CE6881K, CE5881 and CE6881 support this function.

Step 5 (Optional) Run local-preference least active-linknumber linknumber

The minimum number of active links for enabling preferential forwarding of local
traffic is configured for the Eth-Trunk.

By default, the minimum number of active links for enabling preferential

forwarding of local traffic is not configured for an Eth-Trunk.

If the number of active links on an Eth-Trunk is smaller than the value of

linknumber, the system automatically disables preferential forwarding of local
traffic on the Eth-Trunk and preferential forwarding of local traffic on the chip.

Step 6 Run commit

The configuration is committed.

----End

3.10 Creating an Eth-Trunk Layer 3 Sub-interface

Context
Layer 3 Sub-interfaces can be configured on a Layer 3 Eth-Trunk. When Layer 3
devices connect to Layer 2 devices in different VLANs through the Layer 3 Eth-
Trunk, Layer 3 sub-interfaces must be configured on the Eth-Trunk to identify
packets from different VLANs and to enable users in these VLANs to communicate
with each other.

NOTE

The CE9860EI does not support this function.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 151

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run vlan reserved for l3sub-interface
Reserved VLANs are configured for a Layer 3 sub-interfaces.
NOTE

This command is supported only on the following: CE6857EI, CE6857E, CE6865EI, CE8850EI,
CE8861EI, CE8868EI.

Step 3 Run interface eth-trunk trunk-id

An Eth-Trunk is created and the Eth-Trunk interface view is displayed.
Step 4 Run undo portswitch
A Layer 3 Eth-Trunk is configured.
Step 5 Run quit
Exit the Eth-Trunk interface view.
Step 6 Run interface eth-trunk trunk-id.subnumber
An Eth-Trunk Layer 3 sub-interface is created, and the Layer 3 sub-interface view
is displayed.
subnumber specifies the number of the Layer 3 sub-interface. The value is in the
range 1 to 99999999.

NOTE

By default, the LinkDown alarm (Trap OID: 1.3.6.1.6.1.1.5.3) is generated when the status of a
Layer 2 or Layer 3 sub-interface changes. If a large number of Layer 2 or Layer 3 sub-interfaces
exist on a device, the LinkDown alarm is reported on the sub-interfaces at the interval of several
minutes. In this case, an NMS has to process a large number of interface status change alarms,
which overloads the NMS. To resolve this problem, run the subinterface trap updown disable
command in the system view to disable LinkDown alarm generation on the Layer 2 or Layer 3
sub-interfaces as needed. After this command is run, the LinkDown alarm is no longer
generated on any of the device's Layer 2 or Layer 3 sub-interfaces in case of a status change.
Therefore, exercise caution when running this command.

Step 7 Run ip address ip-address { mask | mask-length } [ sub ]

An IP address is configured for the Layer 3 sub-interface.
When configuring multiple IP addresses for an Eth-Trunk Layer 3 sub-interface, use
the sub keyword to indicate the second and subsequent IP addresses.
Step 8 Run dot1q termination vid vid
The Eth-trunk Layer 3 sub-interface is configured to terminate single-tagged
packets.
Step 9 Run commit
The configuration is committed.

----End

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 152

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3.11 Maintaining Link Aggregation

3.11.1 Enabling LACP Alarm Control

To prevent a device from frequently reporting LACP alarms, LACP alarm control
can be enabled. After this function is enabled, the device reports alarms only when
LACP negotiation fails due to specific reasons.

Prerequisites
Eth-Trunk interfaces have been configured to work in LACP mode.

Context
A device reports an LACP alarm if its Eth-Trunk service in LACP mode fails. To
prevent the device from frequently reporting such alarms, LACP alarm control can
be enabled. After this function is enabled, the device reports
hwLacpNegotiateFailed, hwLacpPartialLinkLoss, hwLacpTotalLinkLoss, or Eth-Trunk
linkdown alarms only when LACP negotiation fails due to the following reasons:
● The device's physical link goes Down.
● LACP negotiation times out.
● LACP determines that packets are looped back.
● LACP determines that the system ID and port key in the LACPDU from the
peer end on the local port are inconsistent with those from the peer end on
the reference port.

Procedure
Step 1 Run system-view

The system view is displayed.

Step 2 Run lacp alarm-control link-failure

LACP alarm control is enabled.

A device that has reported an hwLacpNegotiateFailed, hwLacpPartialLinkLoss,

hwLacpTotalLinkLoss, or Eth-Trunk linkdown alarm will report a clear alarm if the
following conditions are met:

1. The lacp alarm-control link-failure command is run.

2. The trigger conditions for the reported alarm are beyond the four reasons for
LACP negotiation Down.

Although a clear alarm is reported, the problem triggering the alarm persists.

After the lacp alarm-control link-failure command is run, the

hwLacpNegotiateFailed, hwLacpPartialLinkLoss, hwLacpTotalLinkLoss, and Eth-
Trunk linkdown alarms are not reported except for the preceding four reasons.
Therefore, exercise caution when running this command.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 153

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Step 3 Run commit

The configuration is committed.

----End

3.11.2 Configuring a Rule for Collecting Statistics on Packets

Containing Specified 5-tuple Information
Context
The 5-tuple information includes the source and destination IP addresses, source
and destination port numbers, and protocol type. Traffic transmitted on each
device interface contains different 5-tuple information. If the outbound interface is
an Eth-Trunk or packets have multiple ECMP next hops, you can configure rules
for collecting statistics on packets that contain specified 5-tuple information to
facilitate fault locating and traffic forwarding path identification.

Procedure
Step 1 Run system-view
The system view is displayed.
Step 2 Run port forwarding-path path-id pathnum { src-ip src-ip-data [ srcip-mask-len ]
| dst-ip dst-ip-data [ dstip-mask-len ] | protocol { protocolnum | tcp [ l4-src-port
src-port-data | l4-dst-port dst-port-data ] * | udp [ l4-src-port src-port-data | l4-
dst-port dst-port-data ] * } } * statistics precedence precedencenum
A rule for collecting statistics on packets with specified 5-tuple information is
configured.
Step 3 Run commit
The configuration is committed.

----End

Follow-up Procedure
If traffic flows through the interface, you can use the display port forwarding-
path path-id pathnum statistics command to check statistics on the packets that
contain specified 5-tuple information.

3.11.3 Clearing Statistics

Context

NOTICE

Statistics cannot be restored once being cleared.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 154

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Procedure
● Run the reset lacp statistics eth-trunk [ trunk-id [ interface interface-type
interface-number ] ] command to clear statistics on transmitted and received
LACP packets.
● Run the reset port forwarding-path path-id pathnum statistics command
to clear statistics on the packets that contain the specified 5-tuple
information.

----End

3.11.4 Monitoring the LAG Operating

Context
During routine maintenance, run the following commands in any view to check
the LAG operating status.

Procedure
● Run the display eth-trunk [ trunk-id [ interface interface-type interface-
number | verbose ] | brief ] command to check the Eth-Trunk configuration.
● Run the display lacp statistics eth-trunk [ trunk-id [ interface interface-type
interface-number ] ] command to check the statistics about LACPDUs sent
and received in LACP mode.
● Run the display interface eth-trunk [ trunk-id ] command to check the Eth-
Trunk status.
● Run the display eth-trunk membership trunk-id command to check
information about member interfaces of an Eth-Trunk.

----End

3.11.5 Using Ping to Monitor the Reachability of Layer 3 Eth-

Trunk Member Interfaces

Context
Multiple physical interfaces can bundle into an Eth-Trunk, and these physical
interfaces are Eth-Trunk member interfaces. Each member interface uses a
specified transmission path. The path-specific service parameters, such as delay,
jitter, and packet loss ratio, are also different. Therefore, you cannot determine
which member interface is faulty when the quality of services on an Eth-Trunk
deteriorates. To resolve this problem, perform a ping test to detect each physical
link, facilitating faulty link locating.

NOTE

The ping test applies to scenarios where two devices are directly connected through an Eth-
Trunk.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 155

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Pre-configuration Tasks
Before using ping to monitor the reachability of Layer 3 Eth-Trunk member
interfaces, complete the following task:
● Run the undo portswitch command to configure the Eth-Trunk to work in
Layer 3 mode and configure an IP address for the Layer 3 Eth-Trunk.
NOTE

An Eth-Trunk works in Layer 2 mode by default.

Procedure
Step 1 Enable the receive end to monitor Layer 3 Eth-Trunk member interfaces.
1. Run the system-view command to enter the system view.
2. Run the trunk member-port-inspect command to enable the receive end to
monitor Layer 3 Eth-Trunk member interfaces.
By default, the receive end is disabled from monitoring Layer 3 Eth-Trunk
member interfaces.

NOTE

The trunk member-port-inspect command takes effect for all Layer 3 Eth-Trunks on
a device. To test the connectivity of Eth-Trunks, disable this function after detection of
Eth-Trunk member interfaces is completed. If this function is not disabled, the device
keeps monitoring Eth-Trunk member interfaces, which consumes a lot of system
resources.

Step 2 Enable the transmit end to monitor Layer 3 Eth-Trunk member interfaces.
1. Run the ping [ ip ] [ -8021p 8021p-value | -a source-ip-address | -c count | -d
| { -f | ignore-mtu } | -h ttl-value | -i interface-type interface-number | -m
time | -p pattern | -q | -r | -ri | -s packetsize | -system-time | -t timeout | { -
tos tos-value | -dscp dscp-value } | -v | -vpn-instance vpn-instance-name ] *
host [ ip-forwarding ] command to enable the transmit end to monitor the
reachability of Layer 3 Eth-Trunk member interfaces.
NOTE

When testing the reachability of Layer 3 Eth-Trunk member interfaces, you must specify
the -a and -i parameters in the ping command. -a and -i indicate the source IP address
and source interface of ICMP Echo Request packets respectively.

The ping command output contains the following information:

– Response to each ping message: If an Echo Response packet is not
received by the transmit end after the timer expires, the "Request time
out" message is displayed, indicating that an Eth-Trunk member interface
fails. If an Echo Response packet is received, the data bytes, message
sequence number, and response time are displayed, indicating that no
Eth-Trunk member interface fails.
– Final statistics: The statistics include the number of sent and received
packets, percentage of failure response packets, and minimum,
maximum, and average response time.
<HUAWEI> ping -a 192.168.1.1 -i 10ge 1/0/1 10.1.1.2
PING 10.1.1.2: 56 data bytes, press CTRL_C to

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 156

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

break
Reply from 10.1.1.2: bytes=56 Sequence=1 ttl=254 time=2
ms
Reply from 10.1.1.2: bytes=56 Sequence=2 ttl=254 time=1
ms
Reply from 10.1.1.2: bytes=56 Sequence=3 ttl=254 time=2
ms
Reply from 10.1.1.2: bytes=56 Sequence=4 ttl=254 time=1
ms
Reply from 10.1.1.2: bytes=56 Sequence=5 ttl=254 time=2
ms

--- 10.1.1.2 ping statistics ---

5 packet(s) transmitted
5 packet(s) received
0.00% packet loss
round-trip min/avg/max = 1/1/2 ms

----End

3.12 Configuration Examples for Link Aggregation

This section only provides configuration examples for individual features. For
details about multi-feature configuration examples, feature-specific configuration
examples, interoperation examples, protocol or hardware replacement examples,
and industry application examples, see the Typical Configuration Examples.

3.12.1 Example for Configuring Link Aggregation in Manual

Load Balancing Mode (CE Switches Excluding CE6870EI)

Networking Requirements
In Figure 3-18, SwitchA and SwitchB connect to devices in VLAN 10 and VLAN 20
through Ethernet links, and heavy traffic is transmitted between SwitchA and
SwitchB.

SwitchA and SwitchB can provide higher link bandwidth to implement inter-VLAN
communication. Data transmission and link reliability need to be ensured.

Figure 3-18 Link aggregation in manual load balancing mode

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 157

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Configuration Roadmap
The configuration roadmap is as follows:
1. Create an Eth-Trunk and add member interfaces to the Eth-Trunk to increase
link bandwidth.
2. Create VLANs and add interfaces to the VLANs.
3. Configure a load balancing mode to ensure that traffic is load balanced
among Eth-Trunk member interfaces.

Procedure
Step 1 Create an Eth-Trunk on SwitchA and SwitchB, and add member interfaces to the
Eth-Trunk.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchA
[*HUAWEI] commit
[~SwitchA] interface eth-trunk 1
[*SwitchA-Eth-Trunk1] mode manual load-balance
[*SwitchA-Eth-Trunk1] trunkport 10ge 1/0/1 to 1/0/3
[*SwitchA-Eth-Trunk1] commit
[~SwitchA-Eth-Trunk1] quit
<HUAWEI> system-view
[~HUAWEI] sysname SwitchB
[*HUAWEI] commit
[~SwitchB] interface eth-trunk 1
[*SwitchB-Eth-Trunk1] mode manual load-balance
[*SwitchB-Eth-Trunk1] trunkport 10ge 1/0/1 to 1/0/3
[*SwitchB-Eth-Trunk1] commit
[~SwitchB-Eth-Trunk1] quit

Step 2 Create VLANs and add interfaces to the VLANs.

# Create VLAN 10 and VLAN 20, and add interfaces to them. The configuration of
SwitchB is similar to the configuration of SwitchA, and is not mentioned here.
[~SwitchA] vlan batch 10 20
[*SwitchA] interface 10ge 1/0/4
[*SwitchA-10GE1/0/4] port link-type trunk
[*SwitchA-10GE1/0/4] port trunk allow-pass vlan 10
[*SwitchA-10GE1/0/4] quit
[*SwitchA] interface 10ge 1/0/5
[*SwitchA-10GE1/0/5] port link-type trunk
[*SwitchA-10GE1/0/5] port trunk allow-pass vlan 20
[*SwitchA-10GE1/0/5] quit
[*SwitchA] commit

# Configure Eth-Trunk 1 to allow packets from VLAN 10 and VLAN 20 to pass

through. The configuration of SwitchB is similar to the configuration of SwitchA,
and is not mentioned here.
[~SwitchA] interface eth-trunk 1
[~SwitchA-Eth-Trunk1] port link-type trunk
[*SwitchA-Eth-Trunk1] port trunk allow-pass vlan 10 20
[*SwitchA-Eth-Trunk1] commit
[~SwitchA-Eth-Trunk1] quit

Step 3 Configure a load balancing mode for Eth-Trunk 1. The configuration of SwitchB is
similar to the configuration of SwitchA, and is not mentioned here.
[~SwitchA] interface eth-trunk 1
[~SwitchA-Eth-Trunk1] load-balance src-dst-mac
[*SwitchA-Eth-Trunk1] commit
[~SwitchA-Eth-Trunk1] quit

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 158

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

If the device is CE6870EI, see 3.7.8 (Optional) Configuring a Load Balancing Mode
(CE6870EI).

Step 4 Verify the configuration.

Run the display eth-trunk 1 command in any view to check whether the Eth-
Trunk is created and whether member interfaces are added.
[~SwitchA] display eth-trunk 1
Eth-Trunk1's state information is:
Working Mode: Normal Hash Arithmetic: src-dst-mac
Least Active-linknumber: 1 Max Bandwidth-affected-linknumber: 16
Operating Status: up Number of Up Ports in Trunk: 3
--------------------------------------------------------------------------------
PortName Status Weight
10GE1/0/1 Up 1
10GE1/0/2 Up 1
10GE1/0/3 Up 1

The preceding command output shows that Eth-Trunk 1 has three member
interfaces: 10GE1/0/1, 10GE1/0/2, and 10GE1/0/3. The member interfaces are all
in Up state. The Operating Status of Eth-Trunk 1 is up.

----End

Configuration Files
● Configuration file of SwitchA
#
sysname SwitchA
#
vlan batch 10 20
#
interface Eth-Trunk1
port link-type trunk
port trunk allow-pass vlan 10 20
load-balance src-dst-mac
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
interface 10GE1/0/3
eth-trunk 1
#
interface 10GE1/0/4
port link-type trunk
port trunk allow-pass vlan 10
#
interface 10GE1/0/5
port link-type trunk
port trunk allow-pass vlan 20
#
return

● Configuration file of SwitchB

#
sysname SwitchB
#
vlan batch 10 20
#
interface Eth-Trunk1

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 159

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

port link-type trunk

port trunk allow-pass vlan 10 20
load-balance src-dst-mac
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
interface 10GE1/0/3
eth-trunk 1
#
interface 10GE1/0/4
port link-type trunk
port trunk allow-pass vlan 10
#
interface 10GE1/0/5
port link-type trunk
port trunk allow-pass vlan 20
#
return

3.12.2 Example for Configuring Link Aggregation in LACP

Mode

Networking Requirements
To improve bandwidth and connection reliability, configure a link aggregation
group (LAG) on two directly connected Switches, as shown in Figure 3-19. The
requirements are as follows:
● Two active links implement load balancing.
● One link functions as the backup link. When a fault occurs on an active link,
the backup link replaces the faulty link to maintain reliable data transmission.

Figure 3-19 Link aggregation in LACP mode

Configuration Roadmap
The configuration roadmap is as follows:
1. Create an Eth-Trunk and configure the Eth-Trunk to work in LACP mode to
implement link aggregation.
2. Add member interfaces to the Eth-Trunk.
3. Set the LACP system priority and determine the Actor so that the Partner
selects active interfaces based on the Actor interface priority.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 160

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

4. Set the upper threshold for the number of active interfaces to improve
reliability.
5. Set LACP interface priorities and determine active interfaces so that interfaces
with higher priorities are selected as active interfaces.

Procedure
Step 1 Create Eth-Trunk 1 on SwitchA and configure Eth-Trunk 1 to work in LACP mode.
The configuration of SwitchB is similar to the configuration of SwitchA, and is not
mentioned here.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchA
[*HUAWEI] commit
[~SwitchA] interface eth-trunk 1
[*SwitchA-Eth-Trunk1] mode lacp-static
[*SwitchA-Eth-Trunk1] commit
[~SwitchA-Eth-Trunk1] quit

Step 2 Add member interfaces to Eth-Trunk 1 on SwitchA. The configuration of SwitchB is

similar to the configuration of SwitchA, and is not mentioned here.
[~SwitchA] interface 10ge 1/0/1
[~SwitchA-10GE1/0/1] eth-trunk 1
[*SwitchA-10GE1/0/1] quit
[*SwitchA] interface 10ge 1/0/2
[*SwitchA-10GE1/0/2] eth-trunk 1
[*SwitchA-10GE1/0/2] quit
[*SwitchA] interface 10ge 1/0/3
[*SwitchA-10GE1/0/3] eth-trunk 1
[*SwitchA-10GE1/0/3] quit
[*SwitchA] commit

Step 3 Set the system priority on SwitchA to 100 so that SwitchA becomes the Actor.
[~SwitchA] lacp priority 100
[*SwitchA] commit

Step 4 On SwitchA, set the upper threshold for the number of active interfaces to 2.
[~SwitchA] interface eth-trunk 1
[~SwitchA-Eth-Trunk1] lacp max active-linknumber 2
[*SwitchA-Eth-Trunk1] commit
[~SwitchA-Eth-Trunk1] quit

Step 5 Set the LACP interface priority and determine active links on SwitchA.
[~SwitchA] interface 10ge 1/0/1
[~SwitchA-10GE1/0/1] lacp priority 100
[*SwitchA-10GE1/0/1] quit
[*SwitchA] interface 10ge 1/0/2
[*SwitchA-10GE1/0/2] lacp priority 100
[*SwitchA-10GE1/0/2] quit
[*SwitchA] commit

Step 6 Verify the configuration.

# Check information about the Eth-Trunk of the Switches and check whether
negotiation is successful on the link.
[~SwitchA] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Static
Preempt Delay: Disabled Hash Arithmetic: profile default
System Priority: 100 System ID: 0025-9e95-7c31
Least Active-linknumber: 1 Max Active-linknumber: 2
Operating Status: up Number Of Up Ports In Trunk: 2

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 161

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Timeout Period: Slow

--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState Weight
10GE1/0/1 Selected 10GE 100 1 20289 10111100 1
10GE1/0/2 Selected 10GE 100 2 20289 10111100 1
10GE1/0/3 Unselect 10GE 32768 3 20289 10100000 1

Partner:
--------------------------------------------------------------------------------
ActorPortName SysPri SystemID PortPri PortNo PortKey PortState
10GE1/0/1 32768 0025-9e95-7c11 32768 4 20289 10111100
10GE1/0/2 32768 0025-9e95-7c11 32768 5 20289 10111100
10GE1/0/3 32768 0025-9e95-7c11 32768 6 20289 10100000
[~SwitchB] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Static
Preempt Delay: Disabled Hash Arithmetic: profile default
System Priority: 32768 System ID: 0025-9e95-7c11
Least Active-linknumber: 1 Max Active-linknumber: 16
Operating Status: up Number Of Up Ports In Trunk: 2
Timeout Period: Slow
--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState Weight
10GE1/0/1 Selected 10GE 32768 4 20289 10111100 1
10GE1/0/2 Selected 10GE 32768 5 20289 10111100 1
10GE1/0/3 Unselect 10GE 32768 6 20289 10100000 1

Partner:
--------------------------------------------------------------------------------
ActorPortName SysPri SystemID PortPri PortNo PortKey PortState
10GE1/0/1 100 0025-9e95-7c31 100 1 20289 10111100
10GE1/0/2 100 0025-9e95-7c31 100 2 20289 10111100
10GE1/0/3 100 0025-9e95-7c31 32768 3 20289 10100000

The preceding information shows that the LACP system priority of SwitchA is 100,
which is higher than the LACP system priority of SwitchB. Member interfaces
10GE1/0/1 and 10GE1/0/2 become the active interfaces and are in Selected state.
Interface 10GE1/0/3 is in Unselect state. Two links are active and work in load
balancing mode, and one link is the backup link.

----End

Configuration Files
● Configuration file of SwitchA
#
sysname SwitchA
#
lacp priority 100
#
interface Eth-Trunk1
mode lacp-static
lacp max active-linknumber 2
#
interface 10GE1/0/1
eth-trunk 1
lacp priority 100
#
interface 10GE1/0/2
eth-trunk 1
lacp priority 100
#
interface 10GE1/0/3
eth-trunk 1
#
return

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 162

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

● Configuration file of SwitchB

#
sysname SwitchB
#
interface Eth-Trunk1
mode lacp-static
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
interface 10GE1/0/3
eth-trunk 1
#
return

3.12.3 Example for Configuring an Eth-Trunk to Work in

Dynamic LACP Mode

Networking Requirements
In Figure 3-20, ServerA is directly connected to SwitchA through an Eth-Trunk in
static LACP mode. ServerB (file server) connected to SwitchA stores ServerA's
network adapter configuration. After ServerA restarts, its configuration is lost and
SwitchA must obtain the configuration from ServerB. SwitchA cannot perform
negotiation with ServerA because ServerA has lost its Eth-Trunk configuration. In
addition, SwitchA's Eth-Trunk cannot go Up and therefore is unable to forward
data.
You can configure the Eth-Trunk on SwitchA to work in dynamic LACP mode.
When LACP negotiation fails, SwitchA can forward packets at Layer 2 through
member interfaces. Therefore, ServerA can successfully obtain the configuration
from ServerB. After ServerA obtains the configuration, an Eth-Trunk in LACP mode
is established. Devices at both ends of the Eth-Trunk send LACPDUs for
negotiation.

Figure 3-20 Eth-Trunk in dynamic LACP mode

Configuration Roadmap
The configuration roadmap is as follows:
1. Configure the Eth-Trunk on SwitchA to work in dynamic LACP mode, and add
Ethernet physical interfaces to the Eth-Trunk to implement link aggregation.
2. Set the upper threshold for the number of active interfaces to improve
reliability.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 163

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3. Set LACP interface priorities and determine active interfaces so that interfaces
with higher priorities are selected as active interfaces.

Procedure
Step 1 Configure the Eth-Trunk on SwitchA to work in dynamic LACP mode and add
Ethernet physical interfaces to the Eth-Trunk.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchA
[*HUAWEI] commit
[~SwitchA] interface eth-trunk 1
[*SwitchA-Eth-Trunk1] mode lacp-dynamic
[*SwitchA-Eth-Trunk1] quit
[*SwitchA] interface 10ge 1/0/1
[*SwitchA-10GE1/0/1] eth-trunk 1
[*SwitchA-10GE1/0/1] quit
[*SwitchA] interface 10ge 1/0/2
[*SwitchA-10GE1/0/2] eth-trunk 1
[*SwitchA-10GE1/0/2] quit
[*SwitchA] interface 10ge 1/0/3
[*SwitchA-10GE1/0/3] eth-trunk 1
[*SwitchA-10GE1/0/3] quit
[*SwitchA] commit

Step 2 On SwitchA, set the upper threshold for the number of active interfaces to 2.
[~SwitchA] interface eth-trunk 1
[~SwitchA-Eth-Trunk1] lacp max active-linknumber 2
[*SwitchA-Eth-Trunk1] quit
[*SwitchA] commit

Step 3 Set LACP interface priorities and determine active links on SwitchA.
[~SwitchA] interface 10ge 1/0/1
[~SwitchA-10GE1/0/1] lacp priority 100
[*SwitchA-10GE1/0/1] quit
[*SwitchA] interface 10ge 1/0/2
[*SwitchA-10GE1/0/2] lacp priority 100
[*SwitchA-10GE1/0/2] quit
[*SwitchA] commit

Step 4 Verify the configuration.

# Run the display eth-trunk command on SwitchA to check Eth-Trunk
information. The command output shows Eth-Trunk information, such as Eth-
Trunk ID (1) and working mode (dynamic LACP mode). 10GE1/0/1, 10GE1/0/2,
and 10GE1/0/3 are in Indep state.
[~SwitchA] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Dynamic
Preempt Delay: Disabled Hash Arithmetic: profile default
System Priority: 32768 System ID: 0025-9e95-7c11
Least Active-linknumber: 1 Max Active-linknumber: 2
Operating Status: up Number Of Up Ports In Trunk: 0
Timeout Period: Slow
--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState
Weight
10GE1/0/1 Indep 10GE 100 0 321 10100010 1
10GE1/0/2 Indep 10GE 100 1 321 10100010 1
10GE1/0/3 Indep 10GE 32768 2 321 10100010
1

Partner:
--------------------------------------------------------------------------------

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 164

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

ActorPortName SysPri SystemID PortPri PortNo PortKey

PortState
10GE1/0/1 0 0000-0000-0000 0 0 0 10100011
10GE1/0/2 0 0000-0000-0000 0 0 0 10100011
10GE1/0/3 0 0000-0000-0000 0 0 0 10100011

# After SwitchA receives LACPDUs from ServerA and link aggregation negotiation
between SwitchA and ServerA succeeds, run the display eth-trunk command on
SwitchA to check Eth-Trunk information. The command output shows Eth-Trunk
information, such as Eth-Trunk ID (1) and working mode (dynamic LACP mode).
10GE1/0/1 and 10GE1/0/2 are active interfaces and in Selected state, and
10GE1/0/3 is in Unselect state.
[~SwitchA] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Dynamic
Preempt Delay: Disabled: Hash Arithmetic: profile default
System Priority: 32768 System ID: 0025-9e95-7c11
Least Active-linknumber: 1 Max Active-linknumber: 2
Operating Status: up Number Of Up Ports In Trunk: 2
Timeout Period: Slow
--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState
Weight
10GE1/0/1 Selected 10GE 100 0 321 10111100 1
10GE1/0/2 Selected 10GE 100 1 321 10111100 1
10GE1/0/3 Unselect 10GE 32768 2 321 10100000
1

Partner:
--------------------------------------------------------------------------------
ActorPortName SysPri SystemID PortPri PortNo PortKey
PortState
10GE1/0/1 32768 0025-9e95-7c31 32768 0 321
10111100
10GE1/0/2 32768 0025-9e95-7c31 32768 1 321
10111100
10GE1/0/3 32768 0025-9e95-7c31 32768 2 321 10100000

----End

Configuration Files
Configuration file of SwitchA
#
sysname SwitchA
#
interface Eth-Trunk1
mode lacp-dynamic
lacp max active-linknumber 2
#
interface 10GE1/0/1
eth-trunk 1
lacp priority 100
#
interface 10GE1/0/2
eth-trunk 1
lacp priority 100
#
interface 10GE1/0/3
eth-trunk 1
#
return

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 165

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

3.12.4 Example for Configuring an Inter-Chassis Eth-Trunk to

Preferentially Forward Traffic Through Local Member
Interfaces

Networking Requirements
As shown in Figure 3-21, SwitchB and SwitchC are connected through stack cables
to increase the total capacity of devices. The two switches functions as a logical
switch. SwitchB functions as the master switch and SwitchC as the backup switch.
To implement backup between devices and improve reliability, physical interfaces
on the two switches are added to an Eth-Trunk. Traffic from two VLANs, VLAN 2
and VLAN 3, is forwarded through both the two interfaces 10GE1/0/1 and
10GE1/0/2 when the network runs properly. This provides high bandwidth use
efficiency between devices and low traffic forwarding efficiency.
To improve traffic forwarding efficiency, each interface should only forward traffic
from one VLAN (in this example, 10GE1/0/1 forwards traffic from VLAN 2 and
10GE1/0/2 forwards traffic from VLAN3). To achieve this goal, configure the Eth-
Trunk to preferentially forward local traffic.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 166

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Figure 3-21 Preferentially forwarding local traffic through the local member
interface

Configuration Roadmap
The configuration roadmap is as follows:
1. Create an Eth-Trunk to implement link aggregation.
2. Add member interfaces to the Eth-Trunk.
3. Enable the Eth-Trunk to preferentially forward local traffic so that traffic is
forwarded by member interfaces on the local device.
4. Configure Layer 2 forwarding to implement Layer 2 connectivity.

Procedure
Step 1 Create an Eth-Trunk and specify the allowed VLANs.
# Configure the stack. In this example, SwitchB is the master switch.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 167

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

<HUAWEI> system-view
[~HUAWEI] sysname SwitchB
[*HUAWEI] commit
[~SwitchB] vlan batch 2 3
[*SwitchB] interface eth-trunk 1
[*SwitchB-Eth-Trunk1] port link-type trunk
[*SwitchB-Eth-Trunk1] port trunk allow-pass vlan 2 3
[*SwitchB-Eth-Trunk1] commit
[~SwitchB-Eth-Trunk1] quit

# Configure SwitchA.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchA
[*HUAWEI] commit
[~SwitchA] vlan batch 2 3
[*SwitchA] interface eth-trunk 1
[*SwitchA-Eth-Trunk1] port link-type trunk
[*SwitchA-Eth-Trunk1] port trunk allow-pass vlan 2 3
[*SwitchA-Eth-Trunk1] commit
[~SwitchA-Eth-Trunk1] quit

Step 2 Add member interfaces to the Eth-Trunk.

# Configure the stack.
[~SwitchB] interface 10GE 1/0/1
[~SwitchB-10GE1/0/1] eth-trunk 1
[*SwitchB-10GE1/0/1] quit
[*SwitchB] interface 10GE 2/0/1
[*SwitchB-10GE2/0/1] eth-trunk 1
[*SwitchB-10GE2/0/1] quit
[*SwitchB] commit

# Configure SwitchA.
[~SwitchA] interface 10GE 1/0/1
[~SwitchA-10GE1/0/1] eth-trunk 1
[*SwitchA-10GE1/0/1] quit
[*SwitchA] interface 10GE 1/0/2
[*SwitchA-10GE1/0/2] eth-trunk 1
[*SwitchA-10GE1/0/2] quit
[*SwitchA] commit

Step 3 In the stack, configure the Eth-Trunk to preferentially forward local traffic.
[~SwitchB] interface eth-trunk 1
[~SwitchB-Eth-Trunk1] undo local-preference disable
[*SwitchB-Eth-Trunk1] commit
[~SwitchB-Eth-Trunk1] quit

Step 4 Configure Layer 2 forwarding.

# Configure the stack.
[~SwitchB] interface 10GE 1/0/2
[~SwitchB-10GE1/0/2] port link-type trunk
[*SwitchB-10GE1/0/2] port trunk allow-pass vlan 2
[*SwitchB-10GE1/0/2] quit
[*SwitchB] interface 10GE 2/0/2
[*SwitchB-10GE2/0/2] port link-type trunk
[*SwitchB-10GE2/0/2] port trunk allow-pass vlan 3
[*SwitchB-10GE2/0/2] quit
[*SwitchB] commit

# Configure SwitchD.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchD
[*HUAWEI] commit

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 168

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

[~SwitchD] vlan 2
[*SwitchD-vlan2] quit
[*SwitchD] interface 10GE 1/0/1
[*SwitchD-10GE1/0/1] port link-type trunk
[*SwitchD-10GE1/0/1] port trunk allow-pass vlan 2
[*SwitchD-10GE1/0/1] quit
[*SwitchD] interface 10GE 1/0/2
[*SwitchD-10GE1/0/2] port link-type trunk
[*SwitchD-10GE1/0/2] port trunk allow-pass vlan 2
[*SwitchD-10GE1/0/2] quit
[*SwitchD] commit

# Configure SwitchE.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchE
[*HUAWEI] commit
[~SwitchE] vlan 3
[*SwitchE-vlan3] quit
[*SwitchE] interface 10GE 1/0/1
[*SwitchE-10GE1/0/1] port link-type trunk
[*SwitchE-10GE1/0/1] port trunk allow-pass vlan 3
[*SwitchE-10GE1/0/1] quit
[*SwitchE] interface 10GE 1/0/2
[*SwitchE-10GE1/0/2] port link-type trunk
[*SwitchE-10GE1/0/2] port trunk allow-pass vlan 3
[*SwitchE-10GE1/0/2] quit
[*SwitchE] commit

Step 5 Verify the configuration.

After the configuration is complete, run the display eth-trunk membership
command in any view. You can view information about Eth-Trunk member
interfaces.
The display on the stack is used as an example.
[~SwitchB] display eth-trunk membership 1
Trunk ID: 1
Used Status: Valid
Type: Ethernet
Working Mode: Normal
Number Of Ports in Trunk: 2
Number Of Up Ports in Trunk: 2
Operating Status: up

Interface 10GE1/0/1, valid, operate up, weight=1,

Interface 10GE2/0/1, valid, operate up, weight=1,

----End

Configuration Files
● Configuration file of the stack
#
sysname SwitchB
#
vlan batch 2 to 3
#
interface Eth-Trunk1
port link-type trunk
port trunk allow-pass vlan 2 to 3
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
port link-type trunk

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 169

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

port trunk allow-pass vlan 2

#
interface 10GE2/0/1
eth-trunk 1
#
interface 10GE2/0/2
port link-type trunk
port trunk allow-pass vlan 3
#
return

● Configuration file of SwitchA

#
sysname SwitchA
#
vlan batch 2 to 3
#
interface Eth-Trunk1
port link-type trunk
port trunk allow-pass vlan 2 to 3
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
return

● Configuration file of SwitchD

#
sysname SwitchD
#
vlan batch 2
#
interface 10GE1/0/1
port link-type trunk
port trunk allow-pass vlan 2
#
interface 10GE1/0/2
port link-type trunk
port trunk allow-pass vlan 2
#
return

● Configuration file of SwitchE

#
sysname SwitchE
#
vlan batch 3
#
interface 10GE1/0/1
port link-type trunk
port trunk allow-pass vlan 3
#
interface 10GE1/0/2
port link-type trunk
port trunk allow-pass vlan 3
#
return

3.12.5 Example for Configuring Inter-device Link Aggregation

in LACP Mode (Standalone Device)

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 170

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

Networking Requirements
In Figure 3-22, DeviceA is dual-homed to DeviceB and DeviceC. The Eth-Trunk in
LACP mode is deployed on DeviceA, and its member interfaces are connected to
10GE1/0/1 and 10GE1/0/2 of DeviceB and DeviceC. 10GE1/0/1 and 10GE1/0/2 use
the same rate and duplex mode. Traffic needs to be load balanced on the two
devices.

Figure 3-22 Dual-active networking where a device is dual-homed through the

Eth-Trunk in LACP mode

Configuration Roadmap
The configuration roadmap is as follows:
1. Create Eth-Trunk 1 in static LACP mode on DeviceA, DeviceB, and DeviceC and
add member interfaces to Eth-Trunk 1 to implement link aggregation.
2. Configure the same LACP system ID on DeviceB and DeviceC.
3. Configure the same LACP system priority on DeviceB and DeviceC.
4. Configure number of Eth-Trunk member interfaces on DeviceC to increase by
32768 to prevent repeated numbers of Eth-Trunk member interfaces in LACP
mode on DeviceB and DeviceC.

Procedure
Step 1 Create Eth-Trunk 1 in LACP mode on DeviceA, DeviceB, and DeviceC and add
member interfaces to Eth-Trunk 1.
<HUAWEI> system-view
[~HUAWEI] sysname DeviceA
[*HUAWEI] commit
[~DeviceA] interface eth-trunk 1
[*DeviceA-Eth-Trunk1] mode lacp-static
[*DeviceA-Eth-Trunk1] trunkport 10ge 1/0/1 to 1/0/4
[*DeviceA-Eth-Trunk1] commit
[~DeviceA-Eth-Trunk1] quit
<HUAWEI> system-view
[~HUAWEI] sysname DeviceB
[*HUAWEI] commit
[~DeviceB] interface eth-trunk 1

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 171

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

[*DeviceB-Eth-Trunk1] mode lacp-static

[*DeviceB-Eth-Trunk1] trunkport 10ge 1/0/1 to 1/0/2
[*DeviceB-Eth-Trunk1] commit
<HUAWEI> system-view
[~HUAWEI] sysname DeviceC
[*HUAWEI] commit
[~DeviceC] interface eth-trunk 1
[*DeviceC-Eth-Trunk1] mode lacp-static
[*DeviceC-Eth-Trunk1] trunkport 10ge 1/0/1 to 1/0/2
[*DeviceC-Eth-Trunk1] commit

Step 2 Set the LACP system ID on DeviceB and DeviceC to 00e0-cf00-0000.

[~DeviceB-Eth-Trunk1] lacp system-id 00e0-cf00-0000
[*DeviceB-Eth-Trunk1] commit
[~DeviceB-Eth-Trunk1] quit
[~DeviceC-Eth-Trunk1] lacp system-id 00e0-cf00-0000
[*DeviceC-Eth-Trunk1] commit
[~DeviceC-Eth-Trunk1] quit

Step 3 Set the LACP system priority on DeviceB and DeviceC to 100.
[~DeviceB] lacp priority 100
[*DeviceB] commit
[~DeviceC] lacp priority 100
[*DeviceC] commit

Step 4 Configure numbers of Eth-Trunk member interfaces on DeviceC to increase by

32768.
[~DeviceC] interface eth-trunk 1
[~DeviceC-Eth-Trunk1] lacp port-id-extension enable
[*DeviceC-Eth-Trunk1] commit

Step 5 Verify the configuration.

# Check information about the Eth-Trunk of each device and check whether the
negotiation is successful.
[~DeviceA] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Static
Preempt Delay: Disabled Hash Arithmetic: profile default
System Priority: 100 System ID: e468-a356-0cb1
Least Active-linknumber: 1 Max Active-linknumber: 16
Operating Status: up Number Of Up Ports In Trunk: 4
Timeout Period: Slow
--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState Weight
10GE1/0/1 Selected 10GE 32768 3 321 10111100 1
10GE1/0/2 Selected 10GE 32768 1 321 10100010 1
10GE1/0/3 Selected 10GE 32768 4 321 10111100 1
10GE1/0/4 Selected 10GE 32768 2 321 10100010 1

Partner:
--------------------------------------------------------------------------------
ActorPortName SysPri SystemID PortPri PortNo PortKey PortState
10GE1/0/1 100 00e0-cf00-0000 32768 32769 321 10111100
10GE1/0/2 100 00e0-cf00-0000 32768 32770 321 10111100
10GE1/0/1 100 00e0-cf00-0000 32768 4 321 10111100
10GE1/0/2 100 00e0-cf00-0000 32768 5 321 10111100
[~DeviceB] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Static
Preempt Delay: Disabled Hash Arithmetic: profile default
System Priority: 100 System ID: 00e0-cf00-0000
Least Active-linknumber: 1 Max Active-linknumber: 32
Operating Status: up Number Of Up Ports In Trunk: 2
Timeout Period: Slow

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 172

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState Weight
10GE1/0/1 Selected 10GE 32768 4 321 10111100 1
10GE1/0/2 Selected 10GE 32768 5 321 10111100 1

Partner:
--------------------------------------------------------------------------------
ActorPortName SysPri SystemID PortPri PortNo PortKey PortState
10GE1/0/3 100 e468-a356-0cb1 32768 4 321 10111100
10GE1/0/4 100 e468-a356-0cb1 32768 2 321 10100010
[~DeviceC] display eth-trunk 1
Eth-Trunk1's state information is:
Local:
LAG ID: 1 Working Mode: Static
Preempt Delay: Disabled Hash Arithmetic: profile default
System Priority: 100 System ID: 00e0-cf00-0000
Least Active-linknumber: 1 Max Active-linknumber: 16
Operating Status: up Number Of Up Ports In Trunk: 2
Timeout Period: Slow
--------------------------------------------------------------------------------
ActorPortName Status PortType PortPri PortNo PortKey PortState Weight
10GE1/0/1 Selected 10GE 32768 32769 321 10111100 1
10GE1/0/2 Selected 10GE 32768 32770 321 10111100 1

Partner:
--------------------------------------------------------------------------------
ActorPortName SysPri SystemID PortPri PortNo PortKey PortState
10GE1/0/1 100 e468-a356-0cb1 32768 3 321 10111100
10GE1/0/2 100 e468-a356-0cb1 32768 1 321 10100010

The preceding information shows that the value of Operating Status on each
device is up, negotiation of Eth-Trunk 1 is successful. Member interfaces on
DeviceB and DeviceC are active interfaces in Selected state, member interfaces of
DeviceB and DeviceC can load balance traffic. The PortNo parameter on DeviceC
indicates that the numbers of Eth-Trunk member interfaces on DeviceC to increase
by 32768.
----End

Configuration Files
● DeviceA configuration file
#
sysname DeviceA
#
interface Eth-Trunk1
mode lacp-static
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
interface 10GE1/0/3
eth-trunk 1
#
interface 10GE1/0/4
eth-trunk 1
#
return
● DeviceB configuration file
#
sysname DeviceB
#
lacp priority 100

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 173

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

#
interface Eth-Trunk1
mode lacp-static
lacp system-id 00e0-cf00-0000
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
return

● DeviceC configuration file

#
sysname DeviceC
#
lacp priority 100
#
interface Eth-Trunk1
mode lacp-static
lacp system-id 00e0-cf00-0000
lacp port-id-extension enable
#
interface 10GE1/0/1
eth-trunk 1
#
interface 10GE1/0/2
eth-trunk 1
#
return

3.13 Troubleshooting Link Aggregation

3.13.1 Traffic Is Unevenly Load Balanced Among Eth-Trunk

Member Interfaces Because the Load Balancing Mode Is
Incorrect
Fault Description
Traffic is unevenly load balanced among Eth-Trunk member interfaces due to the
incorrect load balancing mode.

Procedure
1. Run the display eth-trunk command to check whether the load balancing
mode of the Eth-Trunk meets networking requirements. For example, source
or destination IP address-based load balancing is not recommended in Layer 2
networking.
2. Run the load-balance command to set an appropriate load balancing mode.

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 174

CloudEngine 9800, 8800, 6800, and 5800 Series
Switches
Configuration Guide - Ethernet Switching 3 Link Aggregation Configuration

NOTE

You can set the load balancing mode based on traffic models. When a parameter of
traffic changes frequently, you can set the load balancing mode based on this
parameter to ensure that the traffic is load balanced evenly. For example, if IP
addresses of packets change frequently, use the load balancing mode based on dst-ip,
src-ip, or src-dst-ip. If MAC addresses of packets change frequently but IP addresses
are fixed, use the load balancing mode based on dst-mac, src-mac, or src-dst-mac.

3.13.2 The Physical Status of the Member Interface Is Up But

the Link Protocol Status Is Down Because Link Aggregation Is
Not Configured on the Remote End

Fault Description
SwitchA is configured with link aggregation, but SwitchB is not configured with
link aggregation. As a result, the physical status of the member interface on
SwitchA is Up but the link protocol status is Down.

Procedure
Step 1 Run the display this command on 10GE1/0/1, 10GE1/0/2, and 10GE1/0/3 of
SwitchB. The three interfaces do no join the Eth-Trunk.
If interfaces join the Eth-Trunk, you can view the following configuration in the
command output.
#
interface 10GE1/0/1
eth-trunk 1
#

Step 2 On SwitchB, configure the same link aggregation mode as SwitchA. For details,
see 3.7 Configuring Link Aggregation in Manual Load Balancing Mode or 3.8
Configuring Link Aggregation in LACP Mode.

----End

Issue 02 (2021-04-20) Copyright © Huawei Technologies Co., Ltd. 175