Re: [nsp] about routing

From: Havard.Eidnes@runit.sintef.no
Date: Wed Feb 04 1998 - 15:49:46 EST


> I agree that terminology is confusing sometimes. How about
> 'route balancing' or 'prefix balancing' for a routing protocols
> which distributes its routes among known equal cost paths. Then
> for the forwarding we can accept 'traffic balancing' or 'load
> balancing' refering to the true distribution of bytes and
> packets.

Well, you can call them what you like for me, but I think that
the terminology in itself doesn't really assist all that much in
furthering the understanding of the separation of the issues.

Just to rephrase what Dorian said:

To do "load balancing" between (typically) "equal-cost" paths you
need to think of a couple of separate things:

 o Does your routing protocol permit the selection of more than
   one parallel paths (next-hops) to a given destination?

 o Can the resulting IP forwarding table (which is normally
   constructed based on the collected routing information from
   allt he routing protocols active on the router) hold more than
   one next-hop address for each given destination? (The answer
   to this is normally "yes" nowadays.)

 o How is the forwarding mechanism on the router making use of
   that IP forwarding table. In some cases (e.g. when using a
   "route cache"), the router will pick a single next-hop for
   each new entry it puts in the "route cache", so unless there
   is a cache invalidation all the traffic to a given destination
   will travel over over same wire to the same next-hop. In
   recent route-cache code the "destination" is generally a
   "prefix", not a host (eh, remember good old times? ;-). In
   other cases, the forwarding engine can be made to do a "per-
   packet" choice of next hop (process switching on old gear and
   apparently CEF switching with new software), since no
   simplified route cache (with only a single next-hop) is
   involved in those cases. I'm sure there can be variations on
   these themes, e.g. source-dest hash, as you mention (anyone
   know how cisco's netflow-based route cache does this?),
   although I can see that anything host-based will tend to scale
   badly in backbone routers.

Regards,

- Håvard



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:15 EDT