Re: [manet] Reactive Protocol Situation

[C Chauvenet <[e-mail]>]

Hi,

Le 31 oct. 2012 à 12:01, Jiazi YI a écrit :

Dear all,

Thanks Joe for briefing what had happened in the last several months.

I really don't want to dig into the history, but as long as someone mentioned "results of years of work"

I take it for my own.
Thank you for participating.

Here is a copy of my arguments against option 2) :

C.C> Option 2), would annihilate years of previous work realized for DYMO in the MANET WG.

I get your comment on that.

C.C> Again, I don't see a reason to discard it with a protocol that popped up in MANET 4 months ago. Moreover, the adoption of LOADng may add some misunderstanding about the intend of such a protocol when readers will look at its history. The initial name signification of "LOADng" is a good example, and the last update of the protocol roughly search & replace "LLN" wording by "MANET" as we can see here : http://tools.ietf.org/rfcdiff?url1=draft-clausen-lln-loadng-05.txt. I think such a confusion will not drive the future of internet in a good direction, as JP mentioned. My fear is that it could create confusion and slow down both LLNs and MANET deployments.

What do you think about the rest of my argumentation ?
My "fear" about the confusion is stressed by the "need for a reactive protocol for some types of MANETs" that you just mentioned.

Regards,

Cédric.

, I think those facts are clear with my short memory of MANET:

1. In IETF 87 Maastricht 2010, DYMO was "parked" because the editors "lost the passion" to continue the work, disregarding the comments made for dymo from the mailing list.

2. After that, LOADng was born to meet real needs of the industry. Great efforts have been invested in the last two years, with hundreds iterations on the draft from ~10 authors, interop test, real implementations. Now with concrete results, interop report, mib document, etc.

3. After the DYMO draft sleeping for two years, the editor of DYMO says sorry for the "long delay", and begin to address the comments.
Plus, the main efforts from the parked dymo-21 to the current revision, is to be "compatible" with LOADng.

Ulrich has made very good arguments on LOADng, with running code wide industry support.

On the other hand, DYMO (AODVv2) is taking the design of LOADng to be "compatible", even when the LOADng authors have explicitly expressed that they don't appreciate that.

Maybe it's because I'm so young and so naive that I still believe "running code is the king" in IETF, I don't get the logic why we should force a sophisticated and active document adapt to the one that has been slept for two years.
btw, I have read the latest DYMO revision in detail. A general comments here for DYMO-23 is that, a proof reading is needed first. There are numerous inconsistency in the document:  terminologies, using nonexistent fields ... even duplicated paragraphs (section 5.5.2).

The arguments that made for DYMO are:

1) DYMO had been the WG document
==> true. Actually, it has been there for long time (and can be WG document even forever), but the history told us that this can't help the document evolving.

2) DYMO is the result of years of work.
==> In the contrast, the current situation of reactive protocol in MANET is the result of years of *NO* work on DYMO. IF DYMO had taken the comments from the WG, when reviews were posted around Maastricht, and evolved rather than be dormant for years, then LOADng had not needed to exist.  But unfortunately, there is no magic time machine to bring us back to 2010.

Therefore, I would support WG chairs'  option 2) replace DYMO with LOADng, and strongly against 1) continue with DYMO: the DYMO editors have clearly shown that they couldn't or wouldn't evolve the specification according to feedback, and as there is an industrial need for a reactive protocol for some types of MANETs, we cannot keep sitting around doing nothing while waiting for a miracle.

best

Jiazi




On Oct 31, 2012, at 12:13 AM, Joseph Macker <[e-mail]<mailto:[e-mail]>> wrote:

Hello MANET working group (form Stan and Joe),

As you are all probably aware, there has been WG activity lately on competing drafts for a MANET reactive protocol - DYMO (reviving the current working group document that was parked due to inactivity), and LOADng. Many months ago there was a somewhat authorship led movement towards a common document effort and given positive feedback at the time we the chairs thought this was the best approach given the authors potential to come together and gain the best of both efforts.  Since that period, there has been some fairly strident and rancorous "at times" debate between the authors of the two documents.

During IETF 84 in Vancouver, the co-chairs held a discussion with some of the co-authors of the two documents. Our guidance to the co-authors was to find a way to merge the two documents into one, as it was perceived that are not technically far apart and they both derive roughly from AODV concepts and LOADng had fairly active authorship and implementation efforts. We provided a co-editing proposal to the authors and gave them the timeframe of the Atlanta to come up with an answer back to us regarding this.  As of this writing, those discussions of a potential commonn document and authorship merger have failed.

Therefore, we find ourselves at a crossroads. The authors of the two documents are divided, and it is unlikely that progress on a merged document can be reached based upon recent author feedback. I have also polled the earlier WG editor of DYMO, Ian Chakeres, and he is somewhat disengaged on the issue at the present time.  We see only 3 possible paths forward:

1. Continue the work on the DYMO document, starting with whether there is consensus on its continued approach and also the desire to rename it to AODVv2.
2. Replace the existing DYMO document effort with the LOADng related document effort, defusing ealier references to LLNs as recommended in the last meeting minutes, and to focus more motivationally on general MANET problem spaces (the authors seem to have agreed to this issue if its a WG document).
3. Remove the working group charter for a reactive protocol, effectively killing both documents, at least from a working group (WG) standpoint. This would not be a reflection on the technology in either case, just an admission that we are not working together and reaching consensus.

The co-chairs request and need your opinions on the options.  We have been some silent collecting initial feedback and waiting for author feedback at this point.  Stan and I are both on travel prior to Atlanta so our responses may be sparse and we will also likely be in a "receive mode" for a few days.  So send your opinions.

-Joe
_______________________________________________
manet mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/manet

_______________________________________________
manet mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/manet