OK, I've caught up on the PILC traffic (all 6 of them) and I think
responding to this particular message from Mark will probably cover
the rest;
> > Unfortunately, it strikes me as being a globing together of
> > several broad (relevant & important) areas of interest/research;
> > the resulting issues/problems straddle the entire stack from link
> > characteristics to application behavior & performance.
>
> In some sense we are globbing together a number of topics. But, why
> is that such a problem? Other WGs take on more than one specific,
> highly focused agenda item. And, I think we have the energy to get
> what we outlined finished.
Focused efforts always have a better chance of completing tasks.
I do hope your estimation of the available energy is accurate, but let's
not forget that the traffic on the PEP mailing list dropped off amazingly
fast last summer (despite a large and apparently vigorous subscriber
list). Jim's recent PEP I-D failed to revive it. Sadly, high
attendance at meetings does not always translate into large amounts of
energy from the attendees.
Energy, however is not my major concern - focusing the scope of resulting
discussions is;
The nature of this WG *will* bring out the usual TCP-NG discussions, and
it really doesn't seem that any discussion of "PEP"s (active network
elements) can avoid discussion on active networks in general *and*
revisiting end2end argument issues. I think these are appropriate topics
for discussion and/or exploration within the correct forum, but is this
it?
Put another way, are active network elements considered to be "advanced
development" items, or research? I know that such beasts are already
deployed in commercial networks but does that change anything? People
do lots of unorthodox things on their stub networks and that is fine
as long as it doesn't pollute the orderly flow of traffic on the
shared backbone networks.
If the time were right for PEPish things to be addressed within the
IETF, there is enough meat to justify a separate WG. Personally, I
personally think they are still research items - but ultimately that
isn't my call.
There are plenty of non-PEP items that could be covered instead
(such as identifying how to "fix" applications so they don't make
poor assumptions about the underlying transport environment, and
signaling or otherwise providing visibility to local conditions).
A bucket of stuff is left over from TCPSAT when it was down-scoped
to a only being characterized as a long-delay problem, I would hope
those items would be appropriate for addressing in a PILC group.
> It would seem to me that a series of focused BOFs would not produce
> a final result that was as consistent. But, I don't know that I
> fully grasp the suggestion. Maybe when you get back in email range
> you could explain that a little more.
It's irrelevant at this point with the WG moving forward, but:
The notional difference is striping several documentation efforts
across N working group sessions or devoting a single BOF session to
an in depth discussion/debate about a specific topics (link
characteristics, local signaling, problematic application assumptions,
etc.) I-Ds come into the BOFs with preliminary discussion over the
mailing list. You can cover more ground more thoroughly and
effectively this way. If any of these BOFs reveal significant that warrant
transition for documentation to development, they can potentially spawn
off their own WGs. I don't see any consistency problems stemming from
such an approach.
Remember, not all topics are yet mature enough to bring into the IETF -
there are other places/methods of evolving them to the point where they
*are* appropriate for bringing into the IETF.
Since there is no tetherless equivalent to the ridiculous LA Times article
that led to the creation of the TCPSAT WG, I'm not really clear as to the
burning need for a documentation only WG for tetherless networks (as it
is a rapidly evolving area). I understand the desire, but that is
different from the need.
In any case, I suppose that it will be interesting to watch.
:o)
My views are probably fringish enough to warrant follow-ups be done
in private e-mail to reduce wasted bandwith.
Good Luck!
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:20 EST