Owner's Manual

LACP

The current LACP standard supports dynamic LAG formation between two LACP-enabled nodes (one-to-

one). The software’s LACP allows multiple nodes in the active fabric network to form a single logical

LACP group. To achieve this, the external LACP node configures the links to the active fabric network as a

LAG with direct reachability among the multi-nodes in the active fabric that forms the logical LACP

group.

The multi-node LACP uses egress port blocking when direct reachability between LACP nodes has been

established. The direct reachability requirement is based on the split horizon rule. This requirement

prevents duplication of flooding, broadcast, and multicast traffic to the member ports of the same LAG.

The software uses the virtual system ID and port ID generated for each LACP group instead of using the

actual MAC address and port ID to represent the multiple links from the multiple nodes as originating

from one logical LACP node. If you use VLT or mVLT topologies, you can use a virtual link trunking

interconnect (VLTi) port.

LACP neighbors are discovered by monitoring LACP packets from active fabric external ports. The

software creates a single LACP packet for each active external port and waits for response packets from

uplink and downlink external ports. The response LACP packet triggers LACP to collect detailed neighbor

information about multi-homed connectivity to the active fabric network, the number of links from the

external LACP node, and the configured priority information. The software builds the LACP LAG using the

eligible discovered member ports and synchronizes the members into the LAG group during the

handshake process. The software configures itself as a preferred system priority so that configuration

information, such as ports excluded from LAGs and changing the status to standby, is recognized by

neighbors. If the member LACP port does not receive any LACP packets, a pruning mechanism activates

to change the LACP-enabled port to a normal port as it assumes the LACP configuration was removed.

NIC Teaming

After the network topology is determined and the NIC teaming information from the orchestration or

hypervisor tool is established, the NIC teaming feature allows the controller to identify the NIC teaming

ports and optimize the traffic flows. The software supports the following NIC teaming topologies:

• Active-Standby — Only ports with an active configuration are used for sending and receiving packets

and standby ports remain in standby status until failover

• LACP — Port initiates dynamic handshake to form LAG and ports with agreed status are used for

sending and receiving packets

Traffic Optimization over LACP and NIC Teaming

Both LACP and NIC teaming may require extending source MAC addresses learned on ports from one

active fabric switch to the other switches in the active fabric. For configurations using both NIC teaming

and active-standby, information is not transmitted to other nodes because only active links are used for

sending or receiving traffic.

LACP Requirements and Limitations

• All members of a LAG or port channel must be the same speed (1G, 10G, or 40G). Combinations of

different link speeds are not supported.

• Up to 16 ports are supported for a single node.

• You must directly connect each node in the multi-node LACP to another node. If you have not

established full mesh networks over the nodes, LACP does not form a single LACP group.

Supported Technologies and Protocols