Re: Congestion issue [Re: shim6-proto-07 review]

[Brian E Carpenter <brc@zurich.ibm.com>]

Iljitsch van Beijnum wrote:
> On 14-dec-2006, at 14:31, Brian E Carpenter wrote:
>
> > I am a little puzzled by this concern. In a conventional  multihoming 
> > situation
> > where BGP4 routing changes, we get an unplanned and unexpected path  
> > change,
> > and TCP deals with it as best it can. As far as TCP is concerned,  
> > shim6 is
> > the same - an unexpected path change. On the "first do no harm"  
> > principle,
> > shim6 can do exactly the same as BGP4 multihoming - nothing.
>
>
> The scale of shim multihoming is potentially quite different than  that 
> of BGP multihoming that we have today. Presumably, in a current  setup, 
> the amount of backup bandwidth is such that any congestion  after a path 
> change is survivable. On the other hand, the shim allows  many more 
> people to multihome, including single hosts that may have  both very 
> fast and very slow links. For instance, my laptop has a 1  Gbps 
> interface and a 9600 bps interface, five orders of magnitude.  With BGP, 
> I've never even seen two.

Understood, but this won't break TCP will it? So while this
is a longer term concern, I don't see why we need to deal with
it now. We could usefully document that we are not dealing with it.

   Brian