Fast EtherChannel question

Dilan77

Hi

I am having difficulty visualising Fast EtherChannel. According to the books and definitions on the web, it's a bundle of links, treated as one virtual link, to allow for redundancy.

I can't get this picture into my head though! Can you get this on access layer switches, such as the 2950, to distribution layer? Or are they distribution to core only?

Either way, am I right in thinking of it as the following example -

i) A 2950 and 3550 switch are connected together, across fa0/1 to fa0/4 on each switch (i.e. fa0/1 on the 2950 is connected to fa0/1 on the 3550 and so on)

ii) fa0/1 - fa0/4 are virtually bundled together as one 'link' between the two switches

iii) In the event that one of the physical links dies, e.g. fa0/2 on 2950 to fa0/2 on 3550, the switches distribute the load over the remaining physical links (fa0/1, fa0/3, fa0/4)

If so, how does this tie in with Spanning Tree, as surely with STP some of these ports handling the links would be in blocking state?

Any help appreciated!

Humper

Hi,

Looks like we are reading very similar things.

EtherChannel can be Fast Ethernet or Gigabit Ethernet. You are right, it is a bundle of links treated as one virtual link -- but do not forget that all run at there original clock speed. I do not know if I would say its primary purpose is for redundancy, I would say more for increasing available bandwidth between the devices. The beauty of EtherChannel is that it acts as one link, and STP treats it as one. If you setup and EtherChannel of 2,4, or 8 bundles STP will have all the ports in Forwarding state. If a one link in the bundle fails the load will be distributed over the remaining links and the MAC addresses associated with that port will be cleared (I am not 100% sure on that).

Dilan77 wrote:

Hi

I am having difficulty visualising Fast EtherChannel. According to the books and definitions on the web, it's a bundle of links, treated as one virtual link, to allow for redundancy.

I can't get this picture into my head though! Can you get this on access layer switches, such as the 2950, to distribution layer? Or are they distribution to core only?

Either way, am I right in thinking of it as the following example -

i) A 2950 and 3550 switch are connected together, across fa0/1 to fa0/4 on each switch (i.e. fa0/1 on the 2950 is connected to fa0/1 on the 3550 and so on)

ii) fa0/1 - fa0/4 are virtually bundled together as one 'link' between the two switches

iii) In the event that one of the physical links dies, e.g. fa0/2 on 2950 to fa0/2 on 3550, the switches distribute the load over the remaining physical links (fa0/1, fa0/3, fa0/4)

If so, how does this tie in with Spanning Tree, as surely with STP some of these ports handling the links would be in blocking state?

Any help appreciated!