[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: old GSE idea

On Thursday, Apr 10, 2003, at 06:41 Europe/London, Peter Tattam wrote:
I wonder if some of it may be relevant to recent discussions. (you might want
to ignore the junk about ARPing for routing id's - that probably won't fly)
The relevant bit is the break up of the address bits. Apologies if it
resembles other drafts.
No need to apologize.

In general, the idea is that there is a bit of space that is free for the
network to rewrite at will because it is not an end-to-end value.
One way of looking at it is that it is a way of cooperating with NAT
by way of a trade-off: sacrifice a part of the address for the network
to adjust, and constrain the amount of work host developers need to do
to accomodate their existence. There is probably a more p.c. way
of putting this, but I am not the department of warm and fuzzy feelings.

Personally, I like this approach because it's obviously workable today.
However, I prefer at this point to consider the option of unifying the
v4 and v6 Internet routing tables for operational reasons, at least for now.
Perry Metzger and I had a friendly (well...) exchange about this on
a public list some time ago (3-4 Dec 1999). The general idea is that
leveraging off v4 experience and expertise is attractive to large operators
and other in-the-core supporting organizations, and is acceptable
to host-side and other at-the-edge folks because it needn't get in the way
thanks to tunneling, 6to4 and the like.

So, rather than maintain a routing system specific to GSE, as in your GSE+,
why not just stuff a v4 address into the v6 one, such that the v4 address
(for example) describes the decapsulation point of an IPv6-in-IPv4 tunnel?

This doesn't seem to be prima facie at variance with your old idea, which
correctly identifies an effective requirement for an "AFI"-like subfield in the
higher-order bits of the v6 address in order to experiment with different
routing and addressing semantics within the existing v6 address and header syntax.

Sean.