Link Aggregation Group

Overview

A Link Aggregation Group (LAG) port can be created by combining two or more physical ports on the same node into one logical port. Each physical interface is called a member interface. Link aggregation increases the bandwidth by load balancing the traffic across the member interfaces. It also provides redundancy; if one interfaces fails, the traffic is distributed among the remaining links.

There are two types of Link Aggregation Group:

Static LAG: These types of LAGs are manually configured by the administrators. All ports that are operationally Up are considered active members of the LAG.
Dynamic LAG: These types of LAGs automatically bundle multiple physical ports by exchanging Link Aggregation Control Protocol (LACP) Protocol Data Units (PDU) between the connected devices.

RUCKUS Edge software load balances traffic across all operational member ports of a LAG using a hash derived from packet headers. These packet headers include Source IP, Destination IP address, and Layer 4 (TCP/UDP) ports.

Requirements

A Link Aggregation Group requires the following:

Each LAG interface requires at least one physical interface as a member link.
For a dynamic LAG, all member interfaces should be of the same speed.

Considerations

When configuring a Link Aggregation Group, keep the following considerations in mind:

Non-PCI passthrough interfaces should not be configured as LAG member ports and are not a supported configuration. LAG is not supported with VMware® ESXi™ NIC teaming.
A LAG port is considered operationally Up when at least one of its member ports is up. Similarly, it is marked as operationally Down when all the member ports are down.
A physical port can be part of only one LAG at any point of time.
All the member interfaces of a LAG should be of the same speed.

Best Practices

This feature has no special recommendations for feature enablement or usage.

Limitations

The LAG port has the following limitations:

Only Dynamic LAGs (LACP - as defined in IEEE Standard 802.3ad) are supported. RUCKUS Edge does not support Static LAGs.
The interfaces should be in the unconfigured state; it is recommended that the interfaces which are going to be part of the LAG should not have any prior configurations.
Modifying the LACP mode and timeout for an existing LAG can trigger LACP negotiation, potentially leading to traffic disruption.
When a LAG interface is created, it uses the MAC address of the first physical port as its interface MAC. If that port is later removed (which serves as the MAC provider for the LAG), the next member port’s MAC address will be selected as the LAG’s MAC address. This transition may cause a brief traffic disruption. It is strongly recommended to avoid removing the port for which the MAC address is currently being used by the LAG. Configuring the LAG’s MAC address is not supported.

Note: If the ports within a LAG is of different speeds after auto-negotiation, there is no check for the operational speed mismatch.

Prerequisites

This feature has no prerequisites for feature enablement or usage.