Re: on the point of mobility & multihoming

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

On 3-mrt-04, at 3:35, <john.loughney@nokia.com> wrote:

> > From today's session, Dave & Geoff discussed multihoming vs. 
> > mobility.  In
> a working group I chair, we've had similar discussions and from a 
> transport
> layer, maintaining a session during a multihoming change and a 
> mobility change can be quite similar.  I second Dave's comment that 
> trying to solve two things at once can be never-ending. He also 
> mentioned that solving two
> problems seperately can lead to incompatible or extremely complex 
> solutions
> that don't interoperate well.

Multiaddress mulithoming and mobility solutions must incorporate the 
following mechanisms:

1. multiplexing in the presence of more than one address pair
2. adding/removing addresses
3. for multihoming: failover
4. for mobility: rendezvous

It would be stupid to have two sets of mechanisms for 1. and 2., as the 
complexity of having to implement them both such that they can work 
reliably in the presence of the other is sure to be much worse than 
simply sharing them between mobility and multihoming. Unfortunately the 
ways in which current MIPv6 does these are sub-par, so we need to have 
something better. The logical conclusion is that the MIP people will 
have to throw their stuff out and use the new mechanisms, but I don't 
think we're ready to have that fight quite yet.