Unanswered Question
Mar 25th, 2009

So when you have redundant trunk links between swithes, and STP had blocked one ports of the links, it is a waste of resources NOT to use EtherChannel. Is this correct? EtherChannel allows you to aggregate the bandwidth of the two links, so in effect, we get more data from one switch to the other. Correct?

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Richard Burts Wed, 03/25/2009 - 12:26

Yes with Etherchannel the Spanning Tree sees only a single entity and all of the links within the Etherchannel are active and carrying traffic with none of them blocked by Spanning Tree.



Joseph W. Doherty Wed, 03/25/2009 - 12:51

Richard already answered your primary question, but BTW, although EtherChannel provides more bandwidth between devices, it's not the same as having a single link with that bandwidth. Reason being, EtherChannel statically "hashes" traffic between its links. I.e. one link can be at 100% while others at 0%.

Joseph W. Doherty Wed, 03/25/2009 - 16:30


EtherChannel load sharing algorithm is very important, since a poor choice (or lack of [equipment] choice) can lead to the situtation I described happening all the time, but even the best possible choice doesn't guarantee it won't happen.

The Achilles' heel (as far as I know) is one flow is always limited to one link. LB to link is always deterministic and doesn't respond to actual utilization.

Only reason I mentioned the issue at all (edit - especially in this particular forum topic), some might initially think EtherChannel always provides x*links improved performance.

However, I'm not trying to advance a "why not" but instead, a better understanding of the technology (which your reference will also provide).


This Discussion