Ether Channel

persona

When do we really need to configure EtherChannel?
Advantages and disadvantages plz.

Find more posts tagged with

Free for TechExams community: Cybersecurity salary guide

Compare cert salaries and plan your next career move

Button

Comments

EdTheLad

If you have multiple parallel links between two switches spanning-tree will disable all but one to avoid loops.With ether-channel, spanning-tree views all the links as one, so you can use multiple links simultaneously.
Disadvantages would be extra delay and overhead.
Advantages ease of configuration compared with other methods, extra load-balancing possibilities.

persona

thx 4 the info mate. Appreciate it.
What do I type to configure it, supposing I have SW1 and SW2 with fa0/11 and fa0/12 as trunk?
TIA

carveone

persona wrote:

thx 4 the info mate. Appreciate it.
What do I type to configure it, supposing I have SW1 and SW2 with fa0/11 and fa0/12 as trunk?
TIA

Don't think this is on CCNA but I didn't think default route redistribution for ospf was on either and yet, there it was!

Bind each channel with channel-group:

SW1:
interface fast 0/11
channel-group 1 mode on
interface fast 0/12
channel-group 1 mode on

Ditto on switch 2.

Verify with "show spanning-tree" and "show interface trunk" - instead of individual ports you'll see "Po1" which stands for the interface "port-channel1". This is the logical interface created when an Etherchannel is built. You are now using both physical paths between the two switches at one time.

HTH,

Conor.

persona

Very good. Thx much.

Ed Rooney

Definitely not on CCNA. This is a BCMSN topic, although I think Lammle touches on it in the CCNA book.

If you work with stackable switches like the 3750's, cross-stack etherchannel is a must. It can provide a high-bandwidth backbone between switch stacks. It also seems to provide some redundancy since I can lose a switch in the stack and still work. Using the interface range command makes configuration a lot quicker.

http://www.cisco.com/en/US/products/hw/switches/ps5023/products_configuration_example09186a00806cb982.shtml

markzab

Just jumping in real quick...

I haven't learned this topic yet so pardon the question. When you configure this technology, from what you guys just wrote it seems to negate STP from shutting down the extra links to prevent loops and shows them all as one. Doesn't that re-insert the problem of loops? Or does etherchannel have a seperate function to stop them while keeping all the links up? Just wondering. Thanks.

emsrescue

They become 1 "virtual port" so there are no loops. I use etherchannel to connect servers to switches using multiple NIC's.

Jon

Ed Rooney

It does not shut down STP. An etherchannel is treated as a single link, so STP won't block any of the links. Now, if you add a second link that is not configured as part of an etherchannel, STP will block the appropriate link.

markzab

Ed Rooney wrote:

It does not shut down STP. An etherchannel is treated as a single link, so STP won't block any of the links. Now, if you add a second link that is not configured as part of an etherchannel, STP will block the appropriate link.

Ok, then what stops loops on those redundant links if they're kept up by etherchannel? Or are you saying that loops don't occur because as far as everyone else knows those 2-3-4 links are seen as one virtual link to the rest of the switches and hosts, therefore no loops would be possible???

tech-airman

markzab wrote:

Ed Rooney wrote:

It does not shut down STP. An etherchannel is treated as a single link, so STP won't block any of the links. Now, if you add a second link that is not configured as part of an etherchannel, STP will block the appropriate link.

Ok, then what stops loops on those redundant links if they're kept up by etherchannel? Or are you saying that loops don't occur because as far as everyone else knows those 2-3-4 links are seen as one virtual link to the rest of the switches and hosts, therefore no loops would be possible???

Disclaimer: This is a "thought experiment" and therefore has not been tested with actual Cisco switches or network sims.

markzab,

Think of STP working WITH etherchannel instead of STP and etherchannel as independent entitites/technologies that are fighting each other. It's still STP that provides the loop free network.

Imagine three ethernet (10Mbps) switches with like 24 ports each. So there are Switch1, Switch2, Switch3. Now, imagine you have three Cat 5 crossover cables to interconnect the switches. Crossover cable 1 connects port 24 on Switch1 to port 1 on Switch 2. Crossover cable 2 connects port 24 on Switch2 to port 1 on Switch3. Crossover cable 3 connects port 24 on Switch3 to port1 on Switch1. Let's say the magic of STP occurs and Switch1 becomes the root switch.

Now let's use another crossover cable and connect port 23 of Switch1 to port 2 of Switch2. Now, without etherchannel, STP runs again and guess what? STP blocks the fourth crossover cable's link in order to prevent loops. The total bandwidth between Switch1 and Switch2 is still 10Mbps. Now, let's include the fourth crossover cable into an etherchannel between Switch1 and Switch2. Now we have a link between Switch1 and Switch2 that's running at 20Mbps. Since the link is treated as a single link, STP doesn't block the fourth crossover cable link because STP thinks the link between Switch1 and Switch2 is a "magical 20Mbps ethernet port." Now, extend that imagination of connecting 10 ethernet ports on Switch1 to 10 ethernet ports on Switch2 using 10 crossover cables into an etherchannel. Guess what? That's a total bandwidth of 10x10Mbps or 100Mbps. So to STP, that 10 port etherchannel seems like a single 100Mbps port/channel or FastEthernet speeds. Since etherchannel requires a minimum of a 100Mbps FastEthernet port/link, let's transplant thisexample to FastEthernet switches.

So now we've got 3 FastEthernet switches with 24 ports each, FastSwitch1, FastSwitch2, and FastSwitch3. Let's assume each switch port is configured to be running at 100Mbps speeds. Let's connect them up with Cat 5 crossover cables as followed:

FastSwitch1
[list:65f3816071]
Port 1 - Port 24 of FastSwitch3
Port 15 - Port 1 of FastSwitch2
Port 16 - Port 2 of FastSwitch2
Port 17 - Port 3 of FastSwitch2
Port 18 - Port 4 of FastSwitch2
Port 19 - Port 5 of FastSwitch2
Port 20 - Port 6 of FastSwitch2
Port 21 - Port 7 of FastSwitch2
Port 22 - Port 8 of FastSwitch2
Port 23 - Port 9 of FastSwitch2
Port 24 - Port 10 of Fastswitch2

[*]FastSwitch2

Port 1 - Port 15 of FastSwitch1
Port 2 - Port 16 of FastSwitch1
Port 3 - Port 17 of FastSwitch1
Port 4 - Port 18 of FastSwitch1
Port 5 - Port 19 of FastSwitch1
Port 6 - Port 20 of FastSwitch1
Port 7 - Port 21 of FastSwitch1
Port 8 - Port 22 of FastSwitch1
Port 9 - Port 23 of FastSwitch1
Port 10 - Port 24 of FastSwitch1
Port 24 - Port 1 of FastSwitch3.

[*]FastSwitch3

Port 1 - Port 24 of FastSwitch1
Port 24 - Port 1 of FastSwitch2

[/list:o:65f3816071]

Let's connect all 10 links between FastSwitch1 and FastSwitch2 into a single Etherchannel. Guess what? Since each link runs at 100Mbps and there are 10 links, the etherchannel would be operating theoretically at 100Mbps x 10 or 1000Mbps or 1 Gigabit speed. What that means to STP is that it thinks that the single etherchannel is a 1 Gigabit port/link/cable. Ok, before people mention the 8 link maximum for an etherchannel, I was just using 10 links to illustrate the point that multiple 100Mbps FastEthernet ports can seem like a single 1 Gigabit port. So since it seems like a single Gigabit port, expecting a possible loop across an etherchannel is like expecting a possible loop within the pairs of twisted cable inside the UTP patch cable, which shouldn't happen. Therefore to STP, the etherchannel "looks" like a single faster port/link so loops are still avoided.

So in summary, switch ports/links that are part of an etherchannel at maximum (since we're assuming all links are up and none have failed) are the combined speeds of the individual links that make up the etherchannel. To STP, the etherchannel "looks" like a single port/cable/link which is why the individual links that make up the etherchannel are not blocked by STP.

As mentioned by others, the actual configuration of EtherChannel may be beyond the scope of the CCNA exam.

I hope this helps.

Source:

Cisco EtherChannel Technology [EtherChannel] - Cisco Systems - http://www.cisco.com/en/US/tech/tk389/tk213/technologies_white_paper09186a0080092944.shtml

markzab

Tech, If you just hand wrote that entire thing including the port chart, thank you for taking so much time to help. If you copied and pasted the chart, still, thank you.

That part that answered me question it seems was that since the channel would look like a 1Gb link to STP there wouldn't be any loops occuring, correct? Sounds simple enough.