[K12OSN] Nic Teaming Nic Bonding thoughts

David H. Barr dhbarr at gozelle.com
Sun Mar 5 23:13:51 UTC 2006


On 3/5/06, cisna-barry at wc235.k12.il.us <cisna-barry at wc235.k12.il.us> wrote:

> Could anyone give their thoughts on, nic bonding /teaming with using K12LTSP?
> Just wondering if this concept in the real world actually nets a network
> much more bandwidth? Was thinking of possibly setting up nic teaming on
> one of our servers if anyone, that has actually used this setup, has found
> noticable gains in network throughput.

I seem to recall from my clustering days that channel bonding can be
useful, up to a point.

http://www.scl.ameslab.gov/Projects/MP_Lite/dox_channel_bonding.html


> We seem to have very decent bandwidth but it doesnt hurt to try some new
> things along the way. I wondered if this would be more headaches than it
> is worth.

That seems to be the study I recall, and basically I remember coming
away with "More than 3 bonded channels is a waste." and "The cost is
somewhat more than GigaE but significantly less than 10gE or Myrinet".
 Hope this helps answer your question, at least somewhat.

-dhbarr.




More information about the K12OSN mailing list