Channel-bonding/VLAN with Scyld
Horatio B. Bogbindero
wyy at cersa.admu.edu.ph
Thu Jul 12 07:44:54 PDT 2001
On Thu, Jul 12, 2001 at 03:36:41PM +0200, Jerker Nyberg wrote (wyy sez):
> Sorry if I've missed something important here, but why isn't it possible
> to use ordinary routing instead of channel bonding to split the traffic
> between interfaces?
> According to http://snafu.freedom.org/linux2.2/docs/ip-cref/ the
> outgoing traffic could be split with "ip route add default scope link
> nexthop dev eth0 nexthop dev eth1" or something like that.
this sounds neat... does 2.4 have something more powerful. like being
able to set this humongus default route to different interfaces with
a smart heuristic that determines if the link is saturated and will
avoid that link to a certain amount of time... hmmm. sounds like a good
open source project to contribute to...
> For incoming traffic I do not know. A similiar route on every node for
> every other node doesn't scale very well with many nodes I guess. Hmmm. Is
> that the reason?
> On Thu, 12 Jul 2001, Niclas Andersson wrote:
> > Hi Mike,
> > I had the same problems 2 years ago but I gave up bonding. It doesn't
> > improve the latency (often more important than BW) and the stability
> > is too low.
> > The problem is that both interfaces on one system use the same
> > MAC-adress. Most switches do not tolerate two interfaces with the same
> > MAC-adress not even on separate VLANs. (There are a few that does but
> > can't remember them name of them now. The problem is that this piece
> > of information may be a little bit tricky to find. Bonding a la Becker
> > isn't that common in networks.)
> > I didn't succeed with bonding on HP2424. The reason may of course be the
> > D-link interfaces w. DEC21143 I used which sometimes have difficulties
> > negotiating.
> > What could work is the CISCO Ethernet channel which even the HP
> > ProCurve has. (Be certain that the OS in the switch is recent
> > enough!) I'm not familiar with how that thing work. I only heard that
> > it looked more or less the same as the becker-bonding...
> > > I sent an email to the list last week regarding configuring our HP
> > > Procurve 4000m switch for channel-bonding. I am still having major
> > > problems!!
> > >
> > > If the OS was configured for channel-bonding without any switch
> > > configuration, I got only 17Mbps :-( When I turned on trunking for
> > > certain ports on the switch, I got about 100Mbps, which was only a
> > > slight improvement from 1 NIC (plus, it was using SA/DA, so there was
> > > no node-to-node bandwidth improvement).
> > >
> > > I got a response suggesting that I should setup 2 VLANs, and have all
> > > eth0's on VLAN-1 and all eth1's on VLAN-2. The responder said that he
> > > can get 190Mbps. The trunking configuration is supposed to be for
> > > switch to switch configurations. He was using an HP 2400 (or was it
> > > 2424?). The manual I have is for both models, so I assume that it
> > > will work with mine as well.
> > >
> > > I am still having major difficulties with getting it to work with
> > > Scyld.
> > Best regards,
> > Niclas
> > --
> > Niclas Andersson E-mail: nican at nsc.liu.se
> > National Supercomputer Centre Phone: +46 13 281464
> > Linkoping University, S-581 83 Linkoping, Sweden Fax: +46 13 282535
> > _______________________________________________
> > Beowulf mailing list, Beowulf at beowulf.org
> > To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
William Emmanuel S. Yu
Ateneo Cervini-Eliazo Networks (ACENT)
email : william.s.yu at ieee.org
web : http://cersa.admu.edu.ph/
phone : 63(2)4266001-5925/5904
I must have slipped a disk -- my pack hurts!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 232 bytes
Desc: not available
More information about the Beowulf