Re: charter

[smd@ebone.net (Sean Doran)]

| Can you please forward me a single message (other than yours) that objected
| to the merits of my proposal? 

Well, personally I think that if we try to do something "portable"
among v4 and v6, we will just slow down v6.  I do not object to this
if it's a consensus opinion shared with people who have a longer
track record of favouring v6 deployment. :-) :-)

I also think that your comment that v4 should be removed from
"[p]roduce a document that describes how multihoming is done today in IPv4"
on the grounds that v4 and v6 multihoming today are all but identical
is wrong on the technicality that they are not all but identical today,
either in terms of what's actually documented and the fact that
there is very little of it actually being done.   However, perhaps
you mean, make this "[p]roduce a document that describes CIDR multihoming",
and stress that current CIDR technology is applicable to just about
ANY address length, and has some scalability problems independent
of address length.

| Also, while I respect your opinion, I really
| wish you would have shared it earlier, before revising the charter.

Thomas [via Randy] is under heavy time-pressure because of the
rules the IESG (including Randy) is supposed to operate under.

| I would agree with wording that requires us to solve the problem for IPv6,
| so long as it does not preclude from the get go mechanisms that are
| applicable to IPv4. 

I don't think the charter precludes this; I would hesitate at suggesting
that we focus exclusively on things which ARE portable, because my gut
feeling is that we won't find much that doesn't devolve into turning
both frame types into "just" host-to-network protocols, not everyone 
likes that idea, and a fundamental goal of (some) operators is making
sure that if someone comes along with a large cheque, v6 will actually
work.

That (other) operators have different agenda [me 2], should not derail
the fairly reasonable goal of providing sound advice on how to multihome
and how to have multihomed customers, to the obvious audiences.

| we may end up with two completely different multihoming mechanisms 

Fine, and if it's BETTER, I will actually reverse my common rants
that v6 does not solve the v4 routing problems.   8+8 *may* be BETTER.
However, that technology is not here yet.  CIDR is.  CIDR at least works,
and has problems that we know and can partially deal with in advance.

| To be absolutely clear, I am NOT proposing a constant IPv4 v. IPv6 merits
| discussion, and I think such discussion should be out of bounds.

I think those will happen anyway.  I haven't promised to shut up
when there is overmarketing, only to be constructive here at solving
a real problem.

| Such sentences are a polite way of saying,  "I disagree.  Go roll your own."

Well, I disagree, too, fwiw.   v6 has a real problem that will break
any native deployment because of IDR issues (partly solved by the IESG
request wrt the arch document) and because the only current resolution
to those IDR issues really must not beg the question of policy wrt
address allocations.

| The IETF was formed out of operational need.  The charter you propose, it
| seems to me, unnecessarily skirts that operational need of the moment.

Well, I think we can look at v6 in isolation while considering how
to make v6 + best existing IDR work.   Once that's done, I likely
will start agreeing with you that this insufficient to address (other) real 
operational needs that are equally unaddressed in v4 CIDR.

In short, rather than fight about all sorts of possible radical solutions,
let's get one known-to-work solution out the door being as apolitical and
pragmatic as possible.  

	Sean.