LAG Configuration Guide

Link Aggregation increases link bandwidth, enables load sharing of traffic and improves link reliability by aggregating multiple physical links together to form a single logical link. As shown below, Device A and Device B are connected by three physical links, which are bundled into one logical link with a maximum bandwidth equal to the total bandwidth of the three physical links. At the same time, the three physical links are backed up by each other, so that when one of the links goes down, the other two links can still work.

Multiple ports are aggregated to form an aggregation group and these ports are called member ports of that aggregation group. Each aggregation group uniquely corresponds to a logical interface, called link aggregation interface. Aggregation group and link aggregation interface are numbered consistently.

The member ports within the aggregation group have the following states.

Link aggregation is divided into two modes: static and dynamic aggregation.

Once configured, the selected/unselected state of the port is not affected by the network environment and is more stable.

It can adjust the selected/unselected status of the port according to the information on the other side and the local side, which is more flexible. To ensure proper use, the aggregation method chosen for both ends of the aggregation group needs to be the same. The dynamic aggregation mode is implemented through the LACP (Link Aggregation Control Protocol) protocol. The LACP protocol, based on the IEEE 802.3ad standard, is a protocol that enables dynamic link aggregation, where devices running this protocol send LACPDU to each other to exchange information about link aggregation.

The LACP operating modes are Active and Passive (or Inactive); Active mode, i.e. actively sending out LACPDU probes; Passive mode, i.e. not actively sending out LACPDUs until they are received. If the LACP mode of member ports in the dynamic aggregation group is Passive and the LACP mode of the opposite end is also Passive, both ends will not be able to send LACPDU; if the LACP mode of either end is Active, both ends will be able to send LACPDU; to ensure normal use o, at least one of the LAG at both ends must be Active.

Depending on the role, LACP priority can be divided into two categories: system priority and port priority. System priority is used to distinguish between the higher and lower priority of the switches at both ends. When one of the two ends has a higher priority, the other end will select the selected port of this end according to the end with the higher priority, thus making the selected ports of the two ends agree; the port priority is used to distinguish the priority of each member port to become the selected port. The lower the priority value, the higher the priority.

This parameter defines the long and short timeout attribute of the aggregation group, which is the timeout of the member ports waiting to receive LACPDU. If the LACP timeout is short, the peer will send LACPDU quickly (1 LACPDU every 1 second); if the LACP timeout is long, the peer will send LACPDU slowly (1 LACPDU every 30 seconds). It should be noted that the communication status changes rapidly in the actual network, in order to ensure the quality of communication, for dynamic aggregation groups, no matter the configuration of long timeout or short timeout, after the completion of link aggregation configuration, if the status of a member port on the local side is down, the other side will immediately sense it and set the corresponding port status to down. For dynamic aggregation mode, the switches on both ends of the aggregated link will automatically negotiate the status of the ports within their respective aggregation groups, and the user only needs to ensure that when the local ports are aggregated together, the peers are also aggregated together accordingly.

In scenarios where the network structure is simple and changes infrequently, or when connected devices (such as older switches or servers) do not support the LACP protocol, static aggregation can be used to manually bind physical ports to achieve bandwidth aggregation and redundancy.

Table 1 Configuring Static Aggregation Groups

When adding a physical interface to an aggregation group, the interface must be in a free state, meaning it has no IP configuration, is not assigned to a VLAN, and is not part of another aggregation group.

Table 2 Configuring Aggregation Group Member Ports

Table 3 Overview of LAG Configuration Tasks

The LAG default setting is shown in the table below.

Table 4 LAG Default Setting

Table 5 Configure Dynamic Aggregation Group

When adding a physical interface to an aggregation group, the interface must be in a free state, meaning it has no IP configuration, is not assigned to a VLAN, and is not part of another aggregation group.

Table 6 Configuring Aggregation Group Member Ports

When an interface is added to a link aggregation, it periodically sends LACP protocol packets to dynamically maintain the status of the link aggregation group. After configuring the fast-rate option with a short timeout, it sends LACP packets every 1 second. In the event of a link failure, the fast-rate mode can detect the problem more quickly, making it suitable for scenarios where high network availability is required. In scenarios where network stability is high and the recovery time for failures is not critical, a long timeout can be used to reduce unnecessary packet overhead.

Table 7 Configure LACP fast-rate timeout

LACP dynamically negotiates link aggregation by exchanging LACP protocol data units (LACPDUs). But in some cases (such as when the peer device does not support LACP, link failure, or configuration mismatch), LACP negotiation may fail. In this case, configuring the skip fallback function can force the aggregation group state to UP, thereby avoiding link failure and traffic interruption, and improving reliability. In addition, in scenarios where the server needs to be installed through PXE, this feature also needs to be enabled.

Table 8 Configure LACP fallback

Table 9 Configure LACP system priority

Table 10 Configure LACP port-priority

The minimum number of active links in LACP is an important configuration parameter in Link Aggregation Control Protocol (LACP), which ensures that the minimum number of active links in the Link Aggregation Group meets the requirements during operation. When the number of active links is lower than this value, the entire aggregation group status will become down, which can ensure that the minimum available bandwidth of the aggregation group meets business requirements and prevent partial link failures from causing overload of remaining link traffic.

Table 11 Minimum number of active LACP links

System id defaults to the interface MAC address, and when manually configured, the configured address takes effect.

Table 12 Configure LACP system ID

Table 13 Configure LACP system ID

LACP graceful Down is an elegant closure mechanism used to avoid traffic interruption when actively closing the LACP aggregation link. It ensures that devices at both ends of the link synchronously perceive the port closure status through protocol level coordination, thereby achieving smooth traffic switching and avoiding packet loss during traffic switching.

Table 14 Configure LACP graceful-down

The default load balancing method for members of an aggregation group is based on a five-tuple hash. CX864P-NT supports two packet-based hash algorithms: round_robin and random.

Table 15 Configure Load Balancing Algorithm for Aggregation Group Members

Table 16 Set the IP Address for LAGIF

Table 17 Configure the MTU for LAGIF

By default, the MAC address of the interface is dynamically assigned by the system or is the same as the MAC address of the switch. This series supports users to reconfigure the MAC of physical interfaces, VLAN interfaces and link aggregated interfaces.

Table 18 Configure MAC Address for LAGIF

Shutting down the aggregation port will cause all selected ports in the group to become unselected ports and the state of all member ports to change to down.

Table 19 Shutdown the LAGIF

Check the current aggregation group configuration status, display the protocol of each aggregation group of the system and ports in it.

Table 20 LAG Display and Maintenance

Abbreviations in table description: A - Active, I - InActive, Up - up, Dw - Down, N/A - not available, S - selected, D - deselected

Example.

In the example, we can see that the third column of Protocol is identified as LACP or LOADBALANCE, when it is "LACP", it means that member ports are added and removed dynamically via lacp, and the addition and deletion of member ports to the chip follows the 802.1ax protocol. When it is "LOADBALANCE", it means static lag configuration, and the addition and deletion of member ports to the chip is directly dependent on manual configuration.

Device A

#Create VLAN10 and add Ethernet2 to this VLAN.

#Create VLAN 20 and add Ethernet3 to this VLAN.

#Create static aggregation port Lag1 and add Ethernet0, 1 to this aggregation group.

#Add Lag1 to VLAN10, 20.

Device B Similar to Device A, the configuration process is skipped.

The VMs in Vlan10 and Vlan20 ping each other separately and can ping through.

Device A and Device B are connected to each other via their respective Ethernet0, 1. Configure a Layer 2 dynamic link aggregation group on Device A and Device B and implement separate interworking of VMs in VLAN 10 and VLAN 20.

Device A #Create VLAN10 and add Ethernet2 to this VLAN.

#Create VLAN 20 and add Ethernet3 to this VLAN.

#Create dynamic aggregation port Lag1 and add Ethernet0, 1 to this aggregation group.

#Add PortChannel0001 to VLANs 10 and 20.

Device B

Similar to Device A, the configuration process is skipped.

The VMs in Vlan10 and Vlan20 ping each other separately and can ping through.

Device A #Create static aggregation port Lag1 and add Ethernet0, 1 to this aggregation group.

#Set the IP for Lag1.

Device B

Similar to Device A, the configuration process is skipped. 4. Verify the configuration.

Device A pinging Device B can ping through.

Device A

#Create dynamic aggregation port Lag1 and add Ethernet0, 1 to this aggregation group.

#Set the IP for Lag1.

Device B

Similar to Device A, the configuration process is skipped. 4. Verify the configuration.

Device A pinging Device B can ping through.