Link Aggregation with EtherChannel

Link Aggregation with EtherChannel

This article describes Link Aggregation with EtherChannel on Cisco Router and the methods used to create an EtherChannel. An EtherChannel can be configured manually or can be negotiated using the Cisco aggregation protocol (PAgP), exclusive to Cisco, or the link aggregation control protocol (LACP), defined in IEEE 802.3ad.

What is Link Aggregation ?

In Image 1, traffic coming from several links (normally, 100 Mb / s or 1000 Mb / s) is added to the access switch and must be sent to the distribution switches. Due to the aggregation of traffic, there must be links with a higher bandwidth between the access and distribution switches.

It may be possible to use faster links (for example, 10 Gb / s) in the aggregate link between the access and distribution layer switches. However, adding faster links is expensive. In addition, as the speed increases in the access links, not even the fastest possible port in the aggregate link is fast enough to add the traffic coming from all the access links.
It is also possible to multiply the amount of physical links between switches to increase the overall speed of switch-to-switch communication. However, STP is enabled by default on switch devices. STP blocks redundant links to avoid routing loops.
For these reasons, the best solution is to implement an EtherChannel configuration .

Cisco EtherChannel

In the beginning, Cisco developed EtherChannel technology as a switch-to-switch LAN technique to group several Fast Ethernet or Gigabit Ethernet ports into a single logical channel . When an EtherChannel is configured, the resulting virtual interface is called the " port channel ". The physical interfaces are grouped into a port channel interface.

EtherChannel advantages

EtherChannel technology has many advantages:

• Most configuration tasks can be performed on the EtherChannel interface instead of on each individual port, which ensures configuration consistency across all links.
• The EtherChannel depends on the existing switch ports . It is not necessary to update the link to a faster and more expensive connection to have more bandwidth.
• Load balancing occurs between the links that are part of the EtherChannel itself . Depending on the hardware platform, one or more load balancing methods can be implemented. These methods include load balancing from the source MAC to the destination MAC or load balancing from the source IP to the destination IP, through physical links.
• EtherChannel creates an aggregation that looks like a single logical link . When there are several EtherChannel groups between two switches, STP can block one of the groups to avoid switching loops. When STP blocks one of the redundant links, it blocks the entire EtherChannel. This blocks all ports that belong to that EtherChannel link. Where there is only a single EtherChannel link, all physical links in the EtherChannel are active, since STP only sees a single (logical) link.
• EtherChannel provides redundancy, since the general link is seen as a single logical connection. In addition, the loss of a physical link within the channel does not create any change in the topology, so it is not necessary to recalculate the expansion tree. Assuming there is at least one physical link present, the EtherChannel remains operational, even if its overall performance decreases due to the loss of a link within the EtherChannel.

EtherChannel Operation

EtherChannel can be implemented by grouping several physical ports into one or more logical EtherChannel links.
The EtherChannel provides a full-duplex bandwidth of up to 800 Mb / s (Fast EtherChannel) or 8 Gb / s (Gigabit EtherChannel) between a switch and another switch or host. At present, each EtherChannel can consist of up to eight Ethernet ports configured in a compatible way.

Implementation Restrictions

The Cisco IOS switch can currently support six EtherChannels . However, as new IOSs develop and platforms change, some cards and platforms can support a greater number of ports within an EtherChannel link, as well as a larger amount of Gigabit EtherChannels.
The concept is the same, regardless of the speeds or the number of links that are involved. When configuring EtherChannel on the switches, consider the limits and specifications of the hardware platform.
The original purpose of EtherChannel is to increase the speed capacity in the aggregate links between the switches. However, the concept spread as EtherChannel technology gained more popularity, and now many servers also support link aggregation with EtherChannel.
EtherChannel creates a one-to-one relationship, that is, an EtherChannel link connects only two devices. An EtherChannel link can be created between two switches or between a server with EtherChannel enabled and a switch. However, traffic cannot be sent to two different switches through the same EtherChannel link .

The configuration of the individual ports that are part of the EtherChannel group must be consistent on both devices. If the physical ports on one side are configured as trunk links, the physical ports on the other side must also be configured as trunk links within the same native VLAN. In addition, all ports on each EtherChannel link must be configured as Layer 2 ports .
Each EtherChannel has a logical port channel interface, as shown in Image 3. The configuration applied to the port channel interface affects all physical interfaces that are assigned to that interface.

Port Aggregation Protocol (PAgP)

EtherChannels can be formed through negotiation with one of two protocols: PAgP or LACP . These protocols allow ports with similar characteristics to form a channel through dynamic negotiation with adjacent switches.

• It is also possible to configure a static or unconditional EtherChannel without PAgP or LACP.

PAgP (Port Aggregation Protocol) is an exclusive Cisco protocol that helps in the automatic creation of EtherChannel links . When an EtherChannel link is configured through PAgP, PAgP packets are sent between ports suitable for EtherChannel to negotiate the formation of a channel. When PAgP identifies compatible Ethernet links, it groups the links into an EtherChannel. The EtherChannel is then added to the expansion tree as a single port.
When enabled, PAgP also manages the EtherChannel. PAgP packets are sent every 30 seconds . PAgP checks the consistency of the configuration and manages the links that are added, as well as the failures between two switches. When an EtherChannel is created, it ensures that all ports have the same type of configuration.

PAgP modes

PAgP helps create the EtherChannel link by detecting the settings on each side and making sure the links are compatible, so that the EtherChannel link can be enabled when necessary. In Image 4, the modes for PAgP are shown.

• On : this mode forces the interface to provide a channel without PAgP. The interfaces configured in the on mode do not exchange PAgP packets.
• Desired PAgP : This PAgP mode places an interface in an active negotiation state in which the interface initiates negotiations with other interfaces when sending PAgP packets.
• Automatic PAgP : This PAgP mode places an interface in a passive negotiation state in which the interface responds to the PAgP packets it receives, but does not initiate PAgP negotiation.

The modes must be compatible on each side. If one side is configured in automatic mode, it is placed in a passive state, waiting for the other side to start the EtherChannel negotiation. Also, if the other side is set to automatic mode, the negotiation never starts and the EtherChannel channel is not formed. If all modes are disabled using the no command or if no mode is configured, then the EtherChannel is disabled.
On mode manually places the interface in an EtherChannel, without any negotiation. It works only if the other side is also set to on mode. If the other side is set to negotiate the parameters through PAgP, no EtherChannel is formed, since the side that is set to on mode does not negotiate.

Link Aggregation Control Protocol (LACP)

LACP (Link Aggregation Control Protocol) is part of an IEEE (802.3ad) specification that allows grouping several physical ports to form a single logical channel . LACP allows a switch to negotiate an automatic group by sending LACP packets to the peer. Performs a function similar to PAgP with Cisco EtherChannel. Because LACP is an IEEE standard, it can be used to facilitate EtherChannels in multi-vendor environments. On Cisco devices, both protocols are supported.
Note : In the beginning, LACP was defined as IEEE 802.3ad. However, LACP is now defined in the most modern IEEE 802.1AX standard for local and metropolitan area networks.

LACP modes

LACP provides the same negotiation benefits as PAgP. LACP helps create the EtherChannel link by detecting the settings on each side and ensuring they are compatible, so that the EtherChannel link can be enabled when necessary. In Image 5, the modes for LACP are shown.

•  On : this mode forces the interface to provide a channel without LACP. The interfaces configured in the on mode do not exchange LACP packets.
• LACP active : This LACP mode places a port in active negotiation state. In this state, the port starts negotiations with other ports by sending LACP packets.
• Passive LACP : This LACP mode places a port in passive negotiation state. In this state, the port responds to the LACP packets it receives, but does not initiate the negotiation of LACP packets.

As with PAgP, the modes must be compatible on both sides for the EtherChannel link to form. The on mode is repeated, as it creates the EtherChannel configuration unconditionally, without dynamic negotiation of PAgP or LACP.