[ale] Bonding sanity check

Greg Clifton gccfof5 at gmail.com
Wed Feb 24 18:24:15 EST 2010


There is also some potential chance/benefit of having a non-blocking
configuration, esp. on (a) busy network(s). I remember UK (Univ of Kentucky)
having a VERY impressive flop count on a smallish (maybe 48 node AMD Athlon)
cluster using 4 port 10/100 NICs in each node and a hurd of 100MB switches
wired in such a way as to eliminate or at least minimize port blocking at SC
'03, I think, in Phoenix. The Prof in charge said it took a small super
computer to compute the optimum cabling scheme. Anyways, I thought it was a
pretty cool way of building a super computing cluster on the cheap and this
was back in the day when Myrinet was king of the hill @ ~ $2k per node for
high speed low latency connectivity.  If interested, they might still have
the write up posted at aggregate.org.

GC

On Wed, Feb 24, 2010 at 5:00 PM, Jim Kinney <jim.kinney at gmail.com> wrote:

> Absolutely! The top reason is for failover insulation. Loose a switch but
> connectivity.
>
>
> On Wed, Feb 24, 2010 at 10:52 AM, Jeff Hubbs <jhubbslist at att.net> wrote:
>
>> Are there reasonable uses of Ethernet bonding that incorporate
>> connecting slave NICs of a given bond device to completely different
>> switches?  I've been looking over
>>
>> http://www.cyberciti.biz/howto/question/static/linux-ethernet-bonding-driver-howto.php
>> for reference.  I have not seen or used bonds that do anything other
>> than connect to the same switch.
>>
>> - Jeff
>> _______________________________________________
>> Ale mailing list
>> Ale at ale.org
>> http://mail.ale.org/mailman/listinfo/ale
>> See JOBS, ANNOUNCE and SCHOOLS lists at
>> http://mail.ale.org/mailman/listinfo
>>
>
>
>
> --
> --
> James P. Kinney III
> Actively in pursuit of Life, Liberty and Happiness
>
>
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.ale.org/pipermail/ale/attachments/20100224/03faeedf/attachment.html 


More information about the Ale mailing list