ATLANTA, GEORGIA, September 26, 1995 - Is SNMPv2 Really Dead? In a word, no. The factors driving demand for SNMPv2 - enhanced security, more efficient exchange of management data - loom larger now than ever before. The process of specifying, implementing, testing, and deploying these critically needed extensions to SNMP will continue at an accelerated pace during the next few months. However, there is a question as to when these efforts will again become a part of a chartered IETF Working Group.

There has been considerable confusion and uncertainty regarding the status and future of version 2 of the Internet-standard management framework based on the Simple Network Management Protocol (SNMPv2).

The purpose of this press release is to summarize and clarify the situation.

Recent Background

The Internet Engineering Task Force (IETF) rechartered the SNMPv2 Working Group in November 1994 to prepare recommendations regarding the advancement of SNMPv2 to Draft Standard status. Although work continued in the interim, the Working Group had officially been dormant since 1993 when it completed its charter with the publication of 12 documents specifying SNMPv2 which were declared to be Proposed Standards by the Internet Engineering Steering Group (IESG), the relevant standards-setting body.

In June 1995, the Working Group completed a ``tune-up'' of the documents after about eight months of effort, multiple meetings, considerable electronic mail discussions, and incorporation of many of the approximately 130 changes which were proposed for consideration.

However, two of the primary authors of SNMPv2 had nagging concerns about the results of the open process and proposed a complete rewrite of the security aspects and administrative framework. The IETF extended the deadline for completion by only about eight weeks to consider their proposal and any other counter-proposals brought forward by other members of the working group. Approximately 15 such proposals were submitted and discussed.

Partial Consensus Reached

Although the allotted time was unreasonably short to consider a complete rewrite, the Working Group was able to reach partial consensus by the deadline of September 15th. There was general agreement on 11 of the documents comprising SNMPv2, but the group was unable to reach consensus on the remaining documents. In particular, no concensus was reached on the security and administrative frameworks, the area of the design which was re-opened. The discussions in this area narrowed on two competing sets of specifications but the group was unable to come to closure in the short time which was allocated. As a result, an alternative administrative framework based on SNMPv1 with no security was recommended in order to salvage the investment in other aspects of SNMPv2.

At the deadline, the consensus of the Working Group was an agreement in principle to:

• recommend publication of the 11 documents on which there is a consensus;
• publish a specification of an interim administrative framework named SNMPv1.5; and
• to immediately continue work on the approximately 200 pages of specifications relating to the security and administrative aspects.

Shortly thereafter, (to the chagrin of many Working Group members) the IETF Area Director for Network Management informed the Working Group that it will be disbanded upon completion of the 11 documents. She stated that the Working Group will be restarted in the 4th quarter of 1996 in spite of the strong concensus of the Working Group to continue without delay. The Area Director also stated that independent work is expected to continue in the interim on the security and administrative framework aspects.

Current Status

The Working Group is currently considering new drafts of the 11 documents to see if they adequately reflect the agreement in principle. By IETF rules, this process requires at least two weeks.

The next step, assuming agreement can be reached on the 11 drafts, is for a recommendation from the Working Group to the IESG to result in a "Last Call" for comments to be issued by the IESG, and the resolution of any comments resulting from the Last Call. The Last Call typically lasts for two weeks. The normal result is the pronouncement of a change in the standardization status, to be followed soon thereafter by the publication of the relevant documents as RFCs.

The Future

The internetworking industry abhors the vacuum resultig from the lack of secure and interoperable SNMPv2 products, and cannot tolerate delay. There is a critical need for SNMPv2 technologgy which incorporates security including authentication and privacy within an appropriate administrative framework which can be configured remotely.

While this mismatch between these internetworking industry requirements and the current Area Director's schedule creates a question as to when these efforts will again become a part of a chartered IETF Working Group, there is no question that work will continue to specify, implement, test, and deploy these critically needed features. The only question is when these efforts will become part of a chartered IETF Working Group. Work will begin before the 4th quarter of 1996.

When the 11 documents currently under consideration are published, a request will be filed for a new Working Group charter. As unbelieveable as it may sound, based on the Area Director's stated schedule, this request will almost certainly be denied.

However, the decision with regard to chartering a Working Group is the purview of the entire IESG, not an individual Area Director; hence, an appeal to that body may be successful. If a Working Group is chartered, efforts to build an industry-standard and IETF standard will continue in that venue.

If an IETF Working Group is not chartered in a timely fashion, the work to build a ``short-term interim standard'' in the industry will begin by those who have worked on the specifications to date and are interested in continuing without delay. The resulting specifications will be submitted for consideration as an IETF standard when the IETF is interested in chartering the work.

Ideally, the standard can be both an IETF standard and an industry standard. However, if only one is possible, it is preferable for it to be a well-accepted industry standard than to repeat the unhappy experience of having an IETF standard which is not accepted by the industry.

In either case, it is expected that representatives from multiple leading vendors in the SNMP industry will work together to define an appropriate specification. This is far preferable than each vendor implementing their own proprietary security mechanisms which would inevitably lead to interoperability problems.

This press release is accompanied by three collateral documents, two of which are reproductions of the original sources:

• Electronic mail message from the chair of the IETF SNMPv2 Working Group (Bob Stewart) to the IETF Area Director for Network Management (Deirdre Kostick) reporting the status of the Working Group at the deadline for consensus.
• Electronic mail message from the Area Director in response.
• A white paper providing additional background.

For further information about SNMPv2, contact John Southwood, Sales and Marketing Manger, SNMP Research, Incorporated, 3001 Kimberlin Heights Road, Knoxville, Tennessee, 37920. Telephone: +1 423 573-1434.

SNMP Research, Incorporated produces a family of network management products based on the Simple Network Management Protocol (SNMP) and other protocols including agent and network management station software for the computer and communications industries. SNMP Research's founder, Dr. Jeff Case, has long been a contributor in the area of SNMP as an author of many of the relevant sepcifications and has written a multitude of articles and papers on the topic. As a result of his efforts, SNMP Research is in a leadership position with respect to defining the standards and constructing implementations based on SNMP. SNMP Research's implementations of SNMP for management stations and agents form the basis for many of today's implementations.

(NOTE: The e-mail below has been reformatted, but is otherwise exactly what was sent to the SNMPv2 Working Group mailing list.)

From: Bob Stewart 
To: kostick@qsun.att.com
Subject: My Reading of Group Consensus
Date: Fri, 15 Sep 1995 19:02:02 -0400

Deirdre,

After a long painful process, here's my reading of SNMPv2 Working Group
concensus with regard to our charter and deliverables.

We could not come to a conclusion on administration and security.  I tried
to measure a preference for USEC or SNMPv2*.  When I proposed SNMPv2* as a
working base I got a few strong dissents and several of what I interpret as
either specific approvals or don't cares.  I did not issue a similar test
the other direction.  It is my opinion that the working group at large does
not have a preference between the two competing positions.

Although there is some dissent, we have a relatively strong consensus that
work on security should continue immediately.  It is not clear whether the
preference is to continue from where we are with this working group or to
start another.  What is clear is that a lot of people do not want a long
delay.

Although there is some dissent, we have a strong consensus for salvaging
what we can of SNMPv2 under the plan originally called SNMPv1.5.  I believe
this includes everything we'd agreed upon before USEC appeared except for
the administration and security documents.  I recently talked to Keith, and
he does not understand the edits.  We have a bit of work to do to settle
the right set of documents and their content.  We shall make a valiant
attempt to accomplish that by our deadline of 20 September.

Respectfully submitted,

        Bob Stewart
        SNMPv2 Working Group Chair

From: Deirdre Kostick 
To: snmpv2@TIS.COM
Subject: SNMPv2 Working Group Status
Date: Sat, 16 Sep 95 00:57:12 EDT


Bob Stewart, 
 
I concur with your assessments that:

1) the SNMPv2 Working Group could not come to conclusion on the
administrative and security aspects of SNMPv2.

2) there IS working group consensus on progressing the
non-administrative framework and non-security protocol related aspects
of SNMPv2.

On the assessment of the consensus to continue work immediately on the
framework and security aspects, I understand the great desire to have a
solution and to complete the job.  However, I do not believe there is
sufficient cause to believe that an immediate continuation of the
effort will lead to a successful outcome.

Instead, here are my observations:

1) There should be a cooling off period.  Relations in the working
group are horrible, especially between the founders of SNMP.  A
significant portion of the mailing list traffic is wasted on personal
attacks or admonishments.

I am concerned that these poor working relations will spill over into
other work efforts in the NM area and into other areas on the IETF.

2 ) The working group has been analyzing streams of proposals for so
long that it has lost a clear, common understanding of its goals and
requirements. A clear statement of requirements is needed; this has
been requested by several members of the community.

3) What has best served the community in the past is implementation and
deployment experience. Given the maturity of the embedded SNMP product
base, the chances for deployment in vendor products is lessened without
the blessing of an IETF standard. However, this conclustion underscores
the need to make the right choices for the next iteration of SNMP.
Several members of the community have requested more time to work on
the framework and security issues, and to do related development work.

4) The ongoing work on SNMPv2 has drained the NM Area of its senior and
developing technical resources.  Other work in the area is coming to a
near standstill as SNMPv2 efforts continue.

Next Steps:
------------

1) Keith McCloghrie will complete the edits on the 11 documents that
specify the non-adminstrative framework and non-security aspects of
SNMPv2 per the schedule (9/20).  These documents will then be forwarded
to the IESG.

2) The SNMPv2 Working Group will then be disbanded.  Note that if a
significant dispute arises on the non-administrative
framework/non-security aspects, the SNMPv2 WG will be disbanded prior
to the completion of these 11 documents.

3) Work on the administrative framework/security aspects of SNMP will
be re-started in 4Q96. The first step will be the formation of a
working group to identify a common set of goals and requirements for
the next iteration of SNMP. This working group will have two purposes:

  a) to serve as a litmus test to verify that there is a willingness to
work together and an ability to reach consensus on the goals/requirements
of the effort.

  b) production of the common goals/requirements to base the review of
WG output.

Even though there will not be a formal working group, it is expected
that some members of the community will be sufficiently interested in
the SNMP administrative framework/security effort that they will
continue their work independently. It is hoped that the results of
these ongoing efforts will enrich the knowledge and experience of the
community, and will be used as input into the next working group
effort.

Thanks to all of you for your continuing hard work,

Deirdre Kostick
IETF NM Area Director

For further information, contact SNMP Research International, Inc.

3001 Kimberlin Heights Road
Knoxville, Tennessee, 37920
Telephone: +1 865 579-3311
Fax: +1 865 579-6565
Email: info@snmp.com