Re: [re-ECN] Conex charter - now in External Review

[John Leslie <[e-mail]>]

Woundy, Richard <[e-mail]> wrote:
> From: ken carlberg [mailto:[e-mail]] 
>> 
>> I quite agree about not discussing the above questions, or even
>> considering them in any formal manner on this list or IESG discussions.
>> I sympathize with the curiosity that stems from those questions, but
>> they are quite out of scope of the IETF in chartering WGs.  But if
>> others feel differently, I would like to see how this same litmus test
>> has been conducted in the past for *all* other BoFs, working groups, and
>> even drafts.  Point being -- that's a very nasty rathole to fall into.
> 
> I *completely* agree.
> 
> I have been associated with the DECADE work, which was just chartered
> this week. The same questions could have been asked for that work, but
> weren't.

   If the IESG agreed to charter ConEx last Thursday, we would expect
to have seen that announced no later than yesterday. Thus, the absence
of any such announcement is a strong hint that they haven't yet
agreed to the charter. It is reasonable to assume they're considering
whether some changes to the charter language would help.

   (Or, we could look at the draft agenda for May 6th...)

   In the past, Lars has asked other IESG members to put their concerns
in writing. It is reasonable to assume Stewart is responding to such
a request. Please recall, Stewart is a new IESG member, and may not be
familiar with all the history we would like him to be -- i.e. let's
cut him some slack, please.

   IMHO, the proper answer to his question is, "It doesn't matter."

   Even if every major ISP had committed to _not_ deploy whatever ConEx
may come up with (which is _way_ beyond any possibility I can imagine),
the work is still worth doing for the benefit of enterprises with more
than one upstream path.

   IMHO, the latest proposed charter covers this eventuality nicely
by the sentence:
" 
" However, the CONEX WG will initially focus on one use case, where
" the end hosts and the network that contains the destination end host
" are CONEX-enabled but other networks need not be.

which, IMHO, requires us to come up with something useful for that
case (which, as I read it, excludes _every_ ISP along the way).

   I'm sure many of us have ideas of what intermediate ISPs might do
with the ConEx information in each IP packet -- I've expressed some
of mine on the list -- but these are not appropriate WG issues until
we finish the tasks in the inital charter. Suffice it to say that we
_must_ define ConEx information in a manner which does not depend on
intermediate systems to do anything at all (beyond dropping packets).

   (I also note that Stewart asked about MPLS forwarding: that's a
quite separate question which may deserve more of an answer than
I have already given...)

--
John Leslie <[e-mail]>