Re: Host-based may be the way to go, but network controls areneccessary

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

On Fri, 22 Nov 2002, Kurt Erik Lindqvist wrote:

> > I'm not saying pre-CIDR was the garden of eden, but today's
> > aggreagation
> > hides information that could have been used for more optimal routing.

> If what you are saying is that a single unaggregated prefix per site
> was a better choice to find the best paths,  I think that is to greatly
> oversimplify the problem.

CIDR was a very good thing to do as we are essentially now solving this
problem 10 years later so it bought us 10 years. However, it does chip
away at the underlying IP architecture. You don't want to re-architect
at the drop of a hat but at some point it becomes inevitable. We're not
even there yet, but we are certainly feeling the pain of all the
previous chipping.

> > You can hang a motor on a sailboat, but that only makes it a sailboat
> > with a motor, not a motorboat. If you have a sailboat and you need a
> > motor, this makes sense. When designing a new boat, not so much.

> Sorry, this analogy doesn't fly. Creating a architecture that will help
> us prevent DDOS or DOS is good, but will not only involve the
> multihoming model. It will require much more than that.

Of course changing the architecture involves much more than just
multihoming. So yes, this is off-topic on this list.