Those of you who participated in the Enterprise Management Summit last
week in Dallas and who heard my keynote speech already know that I
advocate a two pronged approach to making progress toward an IETF
standard and an industry standard for SNMPv2 security, administrative
framework, and remote configuration.

Those of you who were not there and haven't heard the presentation
may wish to use anonymous ftp to retrieve the presentation slides found
in /pub/emsummit.keynote.ps on ftp.snmp.com and review them to get the
details.  (Karl:  shadow them on your web server?)

In that address, I describe the two-pronged approach.  The short summary
is as follows -- consult the slides for more complete information.

The first prong is designed to lead to an IETF standard:

    1.	Formation of a blue ribbon panel chartered to look at the SNMPv2
	standardization activities to:
		a)  assess where we are;

		b)  look at how we got there and make recommendations for
		    how these problems can be avoided in the future;

		c)  formulate a plan for resolving the problems with the
		    current documents so that they can be published; and

		d)  chart a course for moving forward to finish the
		    incomplete work of the SNMPv2 Working Group, including
		    the deferred work items related to
		    1)  SNMPv2 security, administrative framework, and the
			remote configuration thereof; and
		    2)  the Manager-to-Manager MIB.

	I recommended such a blue-ribbon panel the week before last in a
	message to the IESG and IAB in an early response to the "Last Call"
	and found it interesting that others on the IETF mailing list
	independently came to a similar conclusion.  The complete text of
	that message is appended.  We've heard very little so far about
	how the IETF leadership will respond to the recommendations contained
	in that note but everything we've heard so far is supportive of the
	approach.

The second prong is designed to lead to an industry standard.

    2.  Formation of a extended design team of leading industry
	representatives to continue work immediately on the specifications
	and implementations

	To that end, we have begun two initiatives:

	a)  recruited key individuals (with a special focus on
	    representatives of the leading manager station implementations)
	    who are interested in completing the work NOW; and

	b)  scheduled an initial meeting of what is being called the
	    SNMPv2* Extended Design Team.

We are very pleased with our recruitment efforts to date.  We have enough
participation that we have a VERY good chance for getting a specification
deployed as products, which is something that SNMPv2 classic was less than
stellar at.  While we have actively recruited some key individuals, we
desire to have a process which is open to all, which is the main point of
this note -- an invitation.

Consequently, I am writing this note to to invite you to attend the first
meeting of the SNMPv2* Extended Design Team, which is currently scheduled for
9:00 a.m. to 3:00 p.m. on Friday, December 8, 1995 in the Hyatt at Reunion,
in Dallas Texas.  We should be clear that this is not an official IETF
meeting but is merely timed and located to minimize impact on people's
travel schedules and budgets since many of us will need to be in Dallas
earlier that week.

The purpose of the meeting is to make forward progress on the specifications
for the SNMP Version 2 administrative framework, security, and MIB for remote
configuration and administration.  We will be focusing on only these topics
and those closely related to them; discussion of unrelated aspects of the
SNMPv2 specifications is out-of-scope.

The tentative agenda is as follows:

9:00 -  Tutorial on SNMPv2* features and techniques             SNMPv2* Design
10:15   (see if we can agree on what the specifications say)    Team

10:30 - Review of design decisions                              Extended SNMPv2*
Noon                                                            Design Team

Noon -  Lunch -- on your own
1:00

1:00 -  Discussion of suggested changes                         Extended SNMPv2*
2:30    (see if we can agree on what the specs should say)      Design Team

2:30 -  Planning for interoperability demonstration at          all
3:00    Networld+Interop Las Vegas April 2-4

3 pm    Conclusion

A complete list of documents and reading assignments will be forthcoming.

The meeting is free and is it open to all.  Please feel free to share this
message with anyone you believe might be interested in participating.

The meeting room we have rented currently seats 30.  In order to help our
planning process and to insure that we have enough room to accommodate all
who wish to attend, we request the courtesy of an RSVP to rsvp@snmp.com if
you will be able to be at the meeting.  We don't want to exclude anyone for
lack of adequate seating; nor do we want to have an unnecessarily large room.
Thanks in advance for letting us know your plans.

We hope to see you in Dallas.

regards,
jdc
-----------------------------------------------------------------------------
the text of the referenced note follows:

from: case
to:  iesg
cc:  iab
date:  20 Oct 1995

It is my understanding that the IESG has been asked to issue the "Last Call"
on the SNMPv2 document set.  As you enter this phase, please be advised that
(this may not be new news to many of you) the discussions within the 
SNMPv2 Working Group have been contentious since late 1994 and increasingly
so of late.

The Working Group has been the victim of multiple attempts
by various factions to hijack the open process and these attempts have
been at least partially successful to date.

In my opinion and the opinion of others, the documents which you have been
asked to advance to "Last Call" do not reflect the consensus of the
Working Group gained through the open process, crippled or not.  At the
deadline for the Working Group to conclude (which came after a request
to extend the deadline was denied) the Working Group had reached a general
consensus on what the final documents should look like even though the final
text had not been prepared.  However, in a closed process the editor was
apparently given directions on how the final text should be structured.  The
editor apparently followed these directions but the resulting documents of
the closed process are inconsistent with the consensus reached via the open
process.  This inconsistency has been identified in the Working Group, but
when the Working Group began to discuss the discrepancies, the request for
the "Last Call" was issued in spite of the fact that the inconsistencies
remain unresolved and in spite of the fact that the rules expressed in
RFC 1602 have not been followed.  These are just some of the problems with
the documents.

These facts will be documented more fully in the near future.  In the meantime,
I request that the IESG consider the following:

	1.  The IESG appoint a member of the IETF leadership to help guide
	    and direct the resolution of these problems.

	or

	2.  Appoint a "blue ribbon" panel to 
	    a)  assess where we are;

	    b)  look at how we got there and make recommendations for
	        how these problems can be avoided in the future; 

	    c)  formulate a plan for resolving the problems with the
		current documents; and

	    d)  chart a course for moving forward to finish the incomplete
		work of the SNMPv2 Working Group, including the deferred
		work items related to
		1)  SNMPv2  security, administrative framework, and the
		    remote configuration thereof; and
		2)  the Manager-to-Manager MIB.


It would not be wise to assign the current Area Director, D. Kostick, as
the liaison, as she is a large part of the problem, having been openly
labeled by a portion of the Working Group membership as a "spineless puppet"
(presumably of Marshall Rose).  Someone who is objective and removed from
the recent and current controversy would be a better choice.

For those of you who are not familiar with the "rich" history surrounding
SNMP standardization, the use of a "blue ribbon" panel is not new but has
multiple precedents.  Dr. Vint Cerf chaired two such panels in the past
(see below for a quick summary of these two previous panels) and has already
sent private inquiries about how he might be of help to put the network
management effort back into a productive mode.

I strongly suggest that you consider invoking his good offices again.

We are very anxious to see these problems resolved and the documents updated
to reflect the consensus, so that they can be published.  We would like to
do this in a timely fashion so that work on the security aspects can resume
in the IETF in a timely fashion.  In order to keep the noise level down to
a minimum, we have held off on charter requests for new work until the
current document set is published.

We would also like there to be a good understanding of how we got into this
stinking kettle of fish so that similar problems can be avoided in the future.

regards,
jdc


Background on the two previous Ad Hoc Network Management Review Groups

Back when the earth was still warm and molten and the IETF was young as
a standardization body, there were multiple network management efforts:
the Simple Gateway Management Protocol (SGMP), the High Level Entity
Management System and Protocol (HEMS/HEMP), and a work promulgated by
the NETMAN group, CMIP over TCP/IP (CMOT).  There had been multiple
parallel tracks and duplication of effort.

A meeting of leaders in the various network management efforts was held
on February 29, 1988 hosted by the San Diego Supercomputer Center to
discuss these issues and to prepare an appropriate plan for future work.
The result of the meeting was that there would be two protocol
developments -- SGMP would evolve to become SNMP as a "short-term" interim
standard until ISO-based approaches were tractable, and the NETMAN work
would continue.  These two protocol efforts were married to a set of
protocol- independent MIB developments, based, in part on the HEMS/HEMP
work to date.

The meeting results are documented in RFC 1052 and this meeting is
formally referred to as the meeting of the First Ad Hoc Network Management
Review Group, but is popularly known in the vernacular variously as either
the 1st Internet Management Woodshed Meeting, or Nuremberg I (referring
to the first trial of crimes against the working group and mankind).


The second Ad Hoc Network Management Review as held on June 12, 1989 hosted
in the CNRI offices in Reston, Virginia.  The meeting was held because of
difficulties in having a single SMI and a single set of MIB objects be
defined for use in conjunction with two disparate management protocols.
In this meeting, the CMOT advocates lost their franchise on the future
because they had failed to reduce their efforts to code and the SNMP group
was enabled to continue its work without being required to cooperate with
the CMOT advocates.  This second such meeting, known in certain parts as
either the 2nd Internet Management Woodshed Meeting, or Nuremberg II,
is fully documented in RFC 1109.