On Tue, 2009-08-25 at 21:57 -0400, David Backeberg wrote: > On Tue, Aug 25, 2009 at 12:50 PM, John A. Sullivan > III<[email protected]> wrote: > > You don't necessarily need a switch to support it. One can use alb mode > > in Linux on any old switch and it works reasonably well other than for > > some excessive ARP traffic. However, as we found out the hard way when > > building our Nexenta SAN, bonding works very well with many-to-many > > traffic but does very little to boost one-to-one network flows. They > > will all collapse to the same pair of NICs in most scenarios and, in the > > one mode where they do not, packet sequencing issues will reduce the > > bandwidth to much less than the sum of the connections. Take care - > > Your claims make sense for a typical > Machine A has one IP address > Machine B has one IP address > > And there is only one route between A and B. In this scenario, yes, > all calls take same route. > > But what about giving each machine two addresses, two routes. And > halve your calls between the two paths between the same systems. > Doesn't this get around your problem, and allow you a chance to > saturate double the number of interfaces? > > If you have four interfaces (as my new boxes do), lather, rinse, > repeat. Anybody have any reason why spreading the bandwidth across > multiple routes wouldn't get around this problem? <snip> Yes, that's correct and exactly what we did in our SAN environment. There are some issues. You will generally not want them all on the same IP network - the inbound traffic may spread across the four addresses if told to do so but the reply traffic will likely go out the default interface.
If they are four distinct IP networks, it means dividing the end users among the multiple networks. In the case of our SAN, we did it without a router using logical networks on the same physical medium. With iproute2 and secondary routing tables, one can be even more creative. In fact, having many phones going to one Asterisk device will probably work well with bonding because it is many to one and each combination of MAC addresses will be treated as a different traffic stream. However, if I recall, the testing environment was two or three asterisk systems talking to each other, wasn't it? - John -- John A. Sullivan III Open Source Development Corporation +1 207-985-7880 [email protected] http://www.spiritualoutreach.com Making Christianity intelligible to secular society _______________________________________________ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- AstriCon 2009 - October 13 - 15 Phoenix, Arizona Register Now: http://www.astricon.net asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
