Re: draft-ietf-multi6-multihoming-requirements-06.txt

[Iljitsch van Beijnum <iljitsch@muada.com>]

On donderdag, jun 12, 2003, at 08:10 Europe/Amsterdam, Pekka Savola 
wrote:

> > No load balancing whatsoever means having a primary connection and a
> > backup. For most multihomers paying for bandwidth you only get to use
> > less than 1% of the time is a non-starter.
> The real question was about *controlled, fine-grained* load balancing.
> That's entirely different than e.g. advertising identical prefix to two
> ISP's in an identical fashion.
Our anonymous friend seems opposed to any kind of load balancing: "Even 
to give
people some tools seems to lead to abuse."

The "abuse" we see today is mostly due to people wanting to prevent 
certain traffic from taking a certain route as long as there is full 
connectivity rather than trying to balance traffic over two or more 
connections. For the former you quickly arrive at announcing more 
specifics, for the latter this is seldom necessary.

And also: "Who controls the balancing? The source?  The destination?  
Don't say 'both'." This shows the author brings some assumptions to the 
table that don't hold up for load balancing in multihomed networks. 
Sure, if A and B are directly connected over two circuits and A prefers 
to use 1 while B wants to use 2, you have a problem. But with 
multihoming A's choice of the outgoing circuit doesn't automatically 
dictate B's incoming circuit. And even if it did, shifting the traffic 
where the other end doesn't care should be sufficient in most cases.