Pankaj, Krishna, as I stated in my note, it is my
hope that agreement can be reached prior to the January meeting. If that
is the case, then we will have a single agreed upon proposal when we meet in
January. If we come to that meeting with a single proposal that
has been carefully reviewed by a broad coalition of experts, we are more
likely to be able to move forward with few negative surprises. There would
still be much work to be done by the working group, including scrutinizing
the proposal and seeing where it might unnecessarily restrict present and
future improvements to 802.17. This work is in addition to
filling in the many holes and deficiencies which almost all first drafts
have.
If we do not come to the January meeting with
the two sides in agreement, then we will have to decide what to accept and what
to reject. There may be sections of the standard where the
differing proposals could work together and we could accept either
technique. Other sections may force us to make a selection. Any
attempt to create a draft from sections of multiple proposals will, of course,
have to be carefully analyzed and simulated in the following weeks, trying
to uncover problems created in mating together sections that were not developed
in concert with each other. To maintain our schedule, we will have to
tightly control the insertion of new material in the draft after the
January meeting. The major reason for inserting new material will be
to develop text to fill known deficiencies and to correct uncovered
problems. As we develop great new ideas, our first priority for them will
be to try to leave room in the standard for future
improvements.
As appealing as it is to constantly
improve our draft, we must stay schedule driven, and resist the
temptation to be making unnecessary additions or changes once the draft has been
issued.
The concerns that you express in the attached note
should be factored into the decision making process in choosing the basis for
the draft, assuming that we must do so at the January meeting.
Pankaj and Krishna, I am not minimizing the efforts
of those that are not tightly coupled to the groups that have developed the
drafts. However, as Vice Chair of 802.17, with an eye closely on our
published schedule, I must urge everyone to avoid the temptation of continuing
the creative task of invention, defining what the standard will look like.
Rather we must focus on the other equally creative task of getting what we have
to work exceptionally well. We have already announced that the November
'01 meeting would be the last meeting for new proposals. Creativity won't
be cut off, but should be channeled to fixing what we have, or to future
enhancements of the standard.
Best regards,
Robert D. Love Chair, Resilient Packet Ring Alliance President, LAN
Connect Consultants 7105 Leveret Circle Raleigh, NC
27615 Phone: 919 848-6773 Mobile: 919
810-7816 email: rdlove@xxxxxxxx
Fax: 208 978-1187
----- Original Message -----
Sent: Friday, December 07, 2001 12:42
PM
Subject: RE: [RPRWG] Reaching
Consensus
Very
well said Pankaj. I agree with the points you have raised.
Krishna
Bob:
Thanks a lot for stressing the issue. May I add that the aim is not to
successfully choose either of two proposals, but to:
- Create a proposal that has best of all proposals, so
good features from different proposals are brought in to create a framework
for further discussions
- Keep the proposal as
independent of internal buffer design details as possible
I wouldn't want to go to January to vote on choosing one proposal - I'd
like to create a proposal out of all the proposals. And I'd like to add
things that are not currently present in any of the proposals. And I'd like
to continually enhance the proposal in subsequent meetings. It would be a
proposal we create, we own, and we become accountable for in case it doesn't
succeed.
Currently even we, just a bunch of people within a closed group, can't
agree on different buffer and fairness schemes and we see lots of holes in
different approaches. It would get worse when it goes to the industry.
As technology improves people would start putting more queues, one per
node and one per class, to give fine-grained QoS/CoS, and to avoid HOL
problems. If RPR freezes specs to the hardware level and doesn't allow use
of advanced techniques in traffic engineering, people will simply bypass it
and move on.
Bob, may I propose that we don't even raise an item in January meeting to
vote on choosing one from many proposals, but to start work on a draft
document based on items from different proposals. Sometimes we may leave
two/three options to start with and complete the draft. Later we can review
the options and coalesce into one.
As long as there is a move to choose one proposal from many proposals,
proponents would continue to stick by their stand and try to win votes. They
will keep showing simulations after simulations ad nauseam, but not move an
inch from where they have been all along :-)
=Pankaj
RDLove wrote:
All, as we get ready for our upcoming
January interim meeting we need to redouble our efforts to bring the three
RPR proposals into agreement. There should be multiple avenues of
communication between the Alladin and Gandalf camps, and additional
efforts to integrate the DVJ proposal. I am concerned that moving
toward agreement is being hampered because too few people are
involved. We need to work
through differences and ideally develop a single proposal going into the
January meeting. My best technical judgment is that the group will
have a much better first draft if there is a single proposal going into
the January meeting, than if we arrive at a decision by voting. The
reason is obvious. Voting at a meeting does not allow the quiet
reflection time that a carefully crafted proposal can have that has been
studied closely over a more significant time period beforehand. In
addition, if we go into the January meeting with a single proposal, we
will have the entire meeting to focus on and potentially improve that
proposal, rather than arguing the merits of competing proposals or
creating a new compromise proposal. For the benefit of all RPR stakeholders I implore everyone
involved in this standards development to study the existing proposals and
seek out ways of bringing them together. Best regards, Robert D. Love
Chair, Resilient Packet Ring
Alliance
President, LAN
Connect Consultants
7105
Leveret Circle Raleigh, NC 27615
Phone: 919
848-6773 Mobile: 919
810-7816
email: rdlove@xxxxxxxx
Fax: 208
978-1187
|