RE: [nsp] load-balancing problem ...

From: F. David Sinn (dsinn@cisco.com)
Date: Thu Nov 30 2000 - 13:20:28 EST


Your second statement is the correct one.

Router A will load balance traffic destined for I to routers B and C in a
50% ratio.

Router A does not consider the two paths out of B to I for it's
load-blancing path.

Router A will only consider the number of equal next-hop's it has to get to
a destination when building the load-blanacing entries.

Router B will load balance the traffic it receives destined to I via D and
E. One caveat being that depending on the version of IOS you are running,
you may see all of the traffic from A destined to I passing through B
following only the link to D or to E, and not balanced between D and E. The
effect is called polarization and has changed so as not to happen in latter
versions of 12.0 and 12.1.

David

-----Original Message-----
From: eduard metz [mailto:e.t.metz@usa.net]
Sent: Wednesday, November 29, 2000 1:49 AM
To: cisco-nsp@puck.nether.net
Subject: [nsp] load-balancing problem ...

Hi all,

I've got a question on how Cisco routers load balance traffic.

Consider a network with 3 equal cost paths from A to I:
1.) A B D G I
2.) A B E G I
3.) A C F H I

Note that the first and second path share the same next-hop. Routing
protocol
is either OSPF or ISIS (link-state), CEF is enabled.

Will each path get (approximately) 1/3 of the traffic, i.e. link AB will bet
2/3 of the traffic and splits this further over BD and BE.
Or is traffic split by each node over the equal cost next-hops, i.e. A sends
50% over AB and 50% over AC, and B splits again on 50/50 basis, effectively
leading to 25% on paths 1 and 2, and 50% on path 3.

Would the loadbalancing behaviour by typical to the Cisco implementation, or
is this specific to the way link-state protocols work?

cheers, Eduard

____________________________________________________________________
Get free email and a permanent address at http://www.netaddress.com/?N=1



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:22 EDT