Configuring Link Aggregation (LACP) for AP Backhaul

LACP provides a means of aggregating multiple Ethernet ports into one logical link, thereby increasing the maximum uplink throughput beyond the limits of a single port.

Some Ruckus APs, including the R610, R710, T610, and T710, can use link aggregation control protocol (LACP) as defined in the 802.1ax (formerly 802.3ad) standard to control the bonding of two 1Gbps physical Ethernet ports together to form a single 2Gbps logical link.

The following APs support LACP:
  • R610
  • R710
  • T610/T610s
  • T710/T710s
Note: The R720 AP does not support LACP at initial release.

Link Aggregation and When To Use It

Beginning with 802.11ac "Wave 2," some higher capacity APs such as the R710 have become capable of passing wireless traffic in excess of 1Gbps. This only occurs in extreme cases, such as during concurrent dual-band operation over the widest possible channels and highest possible modulation and coding scheme (MCS) rates in each band, with all traffic flowing in the same direction. In such cases, a single Gigabit Ethernet backhaul will saturate and limit the AP's capacity to less than 1Gbps.

To alleviate this backhaul limitation, link aggregation can be used to bond multiple Gigabit Ethernet links into a single, high capacity logical link. The AP's link partner, typically an Ethernet switch, must also support LACP and be configured to use this feature.

Wi-Fi client capabilities and data demands should be considered before deploying LACP for AP backhaul. For the vast majority of enterprise WLANs, single Gigabit backhaul for APs is more than sufficient.

Configuring LACP

LACP is configurable via the AP CLI. The set bond command is used with the following syntax:
set bond <profile> {options}
   +++++++++++++++++++++++++++++++++++++++++++++++++++++
     ** <profile>: bond0, ...
     ** options:
         - lacp-rate [0,1], 0 for slow, 1 for fast
         - xmit-hash [0,1,2], 0 for L(ayer2), 1 for L3+4, 2 for L2+3
         - {add|delete} <ethX>
   +++++++++++++++++++++++++++++++++++++++++++++++++++++

Profile (profile) – “bond0” is the only valid profile name.

Options:

  • LACP rate (lacp-rate) defines the rate at which the AP asks its link partner (usually the switch) to transmit LACP control packets (LACPDUs). A faster rate allows the link end-points to respond quicker to any changes on the physical interface (for instance, failover in case of one of the ports is disconnected) at the expense of more overhead. The slow (default) rate is adequate for the vast majority of Enterprise WLAN cases.
  • Slow (lacp-rate 0) (default) requests link partner to transmit LACPDUs every 30 seconds.
  • Fast (lacp-rate 1) requests link partner to transmit LACPDUs every one second.
  • Transmit hash (xmit-hash) defines how the AP chooses to distribute packets between the two physical Ethernet links which comprise the bonded link. Network topology and expected traffic flows should be considered when choosing which transmit hash option to use so as to spread traffic as evenly as possible between the two physical links.
  • Layer 2 (xmit-hash 0) (default) uses the source & destination MAC addresses in the packet to determine which physical link the packet is sent over. This is a fully-compliant 802.3ad option.
  • Layer 3 & Layer 4 (xmit-hash 1) uses source & destination IP addresses as well as source & destination ports. This policy uses upper layer protocol information, when available, to generate the hash. This allows packets destined for a particular network peer to be distributed across both physical links, although a single connection is limited to one of the physical links.

    For fragmented packets, layer 4 information is omitted.

    This algorithm is not fully 802.3ad compliant.

  • Layer 2 & Layer 3 (xmit-hash 2) uses source & destination MAC addresses as well as source & destination IP addresses. This algorithm places all traffic to a particular network peer on the same physical link. For non-IP traffic, the formula is the same as for the layer2 transmit hash policy. This policy is intended to provide a more balanced distribution of traffic than layer2 alone, especially in environments where a layer3 gateway device is required to reach most destinations. This algorithm is 802.3ad compliant.
  • Add or Delete (add|delete) are used to explicitly define which physical Ethernet ports (ethx) are part of the bond interface (bond0). At a bare minimum, to enable LACP on the AP this option must be used to add both physical ports to the bond interface.

Examples

Enable with defaults:

:set bond bond0 add eth0
:set bond bond0 add eth1

Enable and modify LACP rate & distribution algorithm:

:set bond bond0 add eth0
:set bond bond0 add eth1
:set bond bond0 xmit-hash 1
:set bond bond0 lacp-rate 1

Apply VLANs untag ID to bonded interface:

:set interface bond0 type trunk untag 777

Caveats

  • Currently, the controller and AP web UIs do not expose LACP settings or configuration. If an AP is configured via AP CLI to bond its Ethernet ports, then any per-Ethernet port setting in the web UIs are ignored by the AP.
  • The 802.3af PoE operating mode of the R710, which limits AP functionality so as to remain within the standard 802.3af PoE power budget, requires that one Ethernet port be disabled. As such, an AP in 802.3af PoE operating mode cannot be configured to bond its Ethernet ports with LACP.
    Note: Refer to the relevant AP detail page for a description of the feature and function limitations imposed under the 802.3af PoE operating mode.
    This means LACP is supported as long as the R710 AP is powered via:
    • 802.3at PoE+
    • Passive PoE injector capable of at least 20W
    • 12V DC power supply
    Note: The switch port must support link layer discovery protocol (LLDP) power over Ethernet/MDI (PoE+) so the R710 can operate in full-power mode.