As you learned in the previous sections, STP essentially works by blocking redundant links between switches or segments. You may wonder why redundant links exist between switches or segments. They exist to provide a backup link in case a link fails. Apart from redundancy, these links can be used to provide extra bandwidth between these switches or segments. Instead of having a single link to transfer the data, traffic can be load balanced between two or more links. Since STP blocks all redundant links, the load balancing capability is lost.

To overcome this limitation of STP, EtherChannel is used to bundle up to 8 links into a single logical link. After the links are bundle, STP only sees a single logical link and is not able to block anything. Etherchannel protocols on the other hand provide methods to eliminate any loops within the physical links while load balancing traffic across them.

Etherchannel protocols also keep track of the status of each physical link. If one of the physical links go down or come back up, the protocols manage the deletion and addition of the link without STP realizing the change. Irrespective of status of physical links, STP only sees a single logical link.

Cisco switches can use the IEEE standard Link Aggregation Control Protocol (LACP) or Cisco’s proprietary Port Aggregation Protocol (PAGP). You have to know that each EthernChannel is called a channel group and a physical port can be added to it by using the channel-group group-number mode on command in the interface configuration mode. An example of this is shown below:

SW1(config)#interface fast 0/11
SW1(config-if)#channel-group 1 mode on

Creating a port-channel interface Port-channel 1

SW1(config-if)#interface fast 0/12
SW1(config-if)#channel-group 1 mode on

SW2(config)#interface fast 0/11
SW2(config-if)#channel-group 1 mode on

Creating a port-channel interface Port-channel 1

SW2(config-if)#interface fast 0/12
SW2(config-if)#channel-group 1 mode on

In the above example, Interfaces fa0/11 and fa0/12 on SW1 and SW2 are combined into an Etherchannel. Now SW1 and SW2 are connected via a single channel group consisting of two physical interfaces. Channel groups logical interfaces are presented as Po<group-number>. For example, the channel group created in the above example will be presented as interface Po1 since channel group number 1 was used. To verify the configuration you can see the output of show interface trunk as shown below:

SW2#show interface trunk
(some output removed for clarity)

Port   Mode        Encapsulation  Status    Native vlan
Po1    desirable  802.1q              trunking         1

To see if the effect of Etherchannel on STP, see the output of show spanning-tree vlan  command as shown below:

SW2#show spanning vlan 10
(some output removed for clarity)

Interface Role   Sts    Cost    Prio.Nbr Type
Po1          Desg  FWD  12      128.65    P2p

Notice that where two physical interface would have shown up, only a single logical interface (Po1) is seen and it is in the forwarding state.

We will now shut down port 0/11 on SW2 and see the effect on the etherchannel and STP

SW2#conf t
SW2(config)#int fast 0/11
SW2(config-if)#shutdown

3w0d: %LINK-5-CHANGED: Interface FastEthernet0/11, changed state to administratively down

SW2#show spanning vlan 10
VLAN0010

Spanning tree enabled protocol ieee

Interface Role   Sts     Cost  Prio.Nbr Type
Po1          Desg  FWD   19    128.65    P2p

SW2#show interface trunk

Port Mode        Encapsulation  Status Native  vlan
Po1  desirable   802.1q              trunking            1

In the above outputs notice that Po1 is still in forwarding mode and the trunk is still active. The status of a physical interface in an Etherchannel does not effect STP or Trunking. This is true as long as a single physical interface remains active in the etherchannel. If all physical interfaces go down, the channel will go down also and effect STP and Trunking.