RE: PI/metro/geo [Re: The state of IPv6 multihoming development]

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

On Fri, 1 Nov 2002, Craig A. Huegen wrote:

> The point that Tony H. is trying to make (I think, correct me if I'm
> wrong here) is that the multi-PA solution makes life very easy for ISP's
> while throwing all of the operational issues of supporting such a solution
> on the end-sites.

Craig, I've been looking at your messages to this list from the past
week or so to see what exactly your position on identifier/locator
solutions are, but I've been unable to distill anything definitive. So
lat me ask you if a solution with the following properties is acceptable
to you:

As far as your own hosts, the DNS and remote hosts are concerned, hosts
in your network have a single address. However, at the edges of your
network you need boxes that do some address magic so your packets get to
pass through backbone networks as if you were using regular PA. Traffic
engineering is still possible as before, or can be done with finer
granularity, at or very close to the edges but requires more effort (ie
extra hardware, or more processing on existing hardware) to do.

You have a choice between being reachable without any multihoming
benefits for people who do not implement the new scheme, but you have to
use PA addresses, or you can use PI but no interoperation with hosts
that don't implement the new multihoming solution(s) or sit behind a box
that handles this for them.

I'm not saying this is exactly what a solution will look like (hopefully
we can improve on this) but I'm pretty confident a more advanced
multi-address solution can meet the above description.

Iljitsch