Re: [nmrg] RE: Proposed IETF Working Group: sming

[Keith McCloghrie <kzm@cisco.com>]

Walter,

Soliciting a set of proposals to try to solve all the problems that
can be identified was the approach that was advocated at the NIM BOF.
The response at the BOF was much scepticism, and comments that the
scope of such an effort was much too broad to have a realistic chance
of success, i.e., while I think there was a consensus on a desire to do
something, there was no consensus to adopt the proposed approach.  In
contrast, several people spoke at the BOF in favour of pursuing the
SMIng work and starting a WG to to standardize it (and I don't recall
anyone speaking against).  Indeed, the work and implementation done in
the IRTF shows that the proposed charter is achievable, in contrast to
the scepticism on the NIM approach.

WGs in the IETF have a much higher success rate when they start from
an existing solution and "tweak" it, as compared to having to choose
between several competing proposals put forward by (competing) proponents.
In fact, it is very hard to get consensus for a choice between several
proposals.  What normally happens is that after spending a year having
each side claiming the benefits of their different approaches, the WG
reaches an impasse, and either the effort comes to an end or multiple
WGs get created, one for each different proposal.  The IMPP WG in the
Apps Area is just the latest example of this.

Regarding your suggestion of determining requirements, I do think that
listing a set of desirable features would be a good stepping stone for
the WG in determining what "tweaks" to SMIng should be considered.

Keith.

> Bert,
> 
> > -----Original Message-----
> > From: Wijnen, Bert (Bert) [mailto:bwijnen@lucent.com]
> > Sent: Friday, October 27, 2000 9:46 AM
> > To: 'sming@ops.ietf.org'
> > Cc: Dan Romascanu
> > Subject: RE: [nmrg] RE: Proposed IETF Working Group: sming
> > 
> > 
> > Here, inline are my responses/comments
> > 
> > > ----------
> > > From: 	Dan Romascanu[SMTP:dromasca@avaya.com]
> > > Sent: 	Friday, October 27, 2000 1:04 PM
> > > To: 	'sming@ops.ietf.org'; Dave Sidor
> > > Cc: 	David Perkins; Andrea Westerinen; mibs; nim
> > > Subject: 	RE: [nmrg] RE: Proposed IETF Working Group: sming
> > > 
> > > 
> > > Bert,
> > > 
> > > I think that Dave Sidor's message is a good example about 
> > some of the not
> > > so
> > > clear issues concerning this new work proposal. Note that I am now
> > > addressing the content, and not the procedure issues.
> > > 
> > Yes, I got that. And I am trying to get a clear statement 
> > that the IESG can
> > make.
> > As you can see, we are all (including, or may be 
> > specifically, ADs) learning
> > as
> > we go.
> > 
> > > 1. You are mentioning that the proposal intents to 'move 
> > SMI forward to
> > > address some of the issues that have been raised in the last so many
> > > years'.
> > > In this case the first step should be to specify which 
> > problems we propose
> > > to solve. Maybe the first item in the deliverables list should be
> > > 'Requirements for SMIv3 document'.
> > That sounds plausible, but we do know quite a few reqmnts already.
> > For sure, I want to get SMI and SPPI back on the same track.
> > Now.. I would assume that the first WG meeting, or this mailing list
> > right now, allows for discussing this topic. Maybe someone can already
> > prepare an I-D to try and list the most important ones.
> >
> You seem to be advocating an evaluation of the NMRG output based on
> undocumented opinions of the problem set we are facing. By ignoring the
> potential set of problems, it is easy to argue for a specific solution and
> to avoid comparing this proposal with alternatives, because the criteria for
> a comparison are undefined.
>  
> > > 2. The text of the proposal still seems to indicate that it 
> > takes upon
> > > itself to provide an answer to the problem of the common information
> > > model.
> > > Dave Sidor read it this way, so did I.  I happen to be 
> > convinced that such
> > > a
> > > model is needed, and deserves a separate framework, and 
> > SMIv3 is not the
> > > answer.
> > I did get this from your message. I think I tried to answer that in my
> > clarifications.
> > And when we do approve the WG, then I will certainly try to 
> > get the wording
> > improved/fixed. Specific text that you think would make it clearer can
> > always
> > be suggested on this list or directly to me.
> > 
> > > 3. I understand so well that we focus on SNMP and COPS, but 
> > it is too
> > > early
> > > for having decided about the solution. I think that I agree 
> > with your
> > > assessment in a previous mail, that the floor should be 
> > open for different
> > > proposals. This would include the good work done in the 
> > ITRF, but should
> > > not
> > > exclude other proposal that may come from different 
> > sources. The future
> > > Charter should have clear text about this.
> > > 
> > The IETF process is open, and at any point people can raise 
> > any issues or
> > concerns they have. And when we start this work, I can also 
> > see that we
> > spend
> > some face to face time to discuss other possible solutions. 
> > We can also have
> > some email discussions on this. But for now... I do want to 
> > strongly suggest
> > that we first take a serious look at the IRTF docs that I 
> > hope will show up
> > rsn.
> > And the charter tries to convey that "strong suggestion". If 
> > the WG decides
> > later on that the "strong suggestion" should be ignored, then 
> > of course the
> > WG chairs and ADs will evaluate the issues raised to see if we need to
> > reconsider.
>
> Given my comments above, I would suggest that it is reasonable to use the
> NMRG specs as input for developing requirements. However to suggest that the
> NMRG specs is the outcome slams the door pretty hard on potential
> alternatives.