Re: OPSAREA - preliminary minutes of the Hiroshima sessionuploaded

Hi Andy, 

> -----Original Message-----
> From: Andy Bierman [mailto:[email protected]] 
> Sent: Wednesday, December 02, 2009 4:31 PM
> To: David Harrington
> Cc: 'Romascanu, Dan (Dan)'; 'OPS Area (E-mail)'
> Subject: Re: [OPS-AREA] OPSAREA - preliminary minutes of the 
> Hiroshima sessionuploaded
> 
> David Harrington wrote:
> > Hi,
> > 
> 
> Sounds interesting.  Sorry I missed the meeting.
> I thought the OPSAWG was just odds and ends that
> would otherwise fall though the cracks and not get done.

I believe that is accurate. 
I think the OPS area needs to drive the discussion of multi-protocol
management and whether YANG should be promoted by the area to other
areas of the IETF.
I don't have a problem with the OPS area deciding that the OPSAWG is a
good place to assign the deliverables for multi-protocol management.
I do have a problem with the odds and ends in OPSAWG driving the
direction of the OPS area.

> 
> Until YANG is published as an RFC, it is not really relevant
> to other WGs besides NETMOD.  The challenge (perhaps for the OPSAWG)
> is to create and execute a coherent development plan for a
> standard framework (IF-MIB, ENTITY-MIB, etc.).
> 
> The NETCONF WG lost its operator focus, and therefore
> lost its operator co-Chair.  The focus was supposed
> to be on configuration, but the notification work
> was done instead.
> 
> I am not convinced that the mere publication of YANG will
> produce any standards-based configuration of
> any real networking devices.  That is going to require
> agreement on core YANG modules that build on existing
> IETF standards, not the ad-hoc reinvention of every
> SNMP or CLI data structure ever written.
> 
> IMO the IESG should have a 5 year plan for success, and it is not
met,
> then the IETF should officially give up on network configuration.

I don't agree with that sentiment.
First I think the IAB should make a recommendation about architectural
directions for network management.
The 2002 workshop did that to a degree, and the result was to
implicitly endorse using multiple protocols.
I think the IETF (not just the IESG) needs to discuss a coherent plan
for moving to a multi-protocol framework.
I think the OPS area should help drive that discussion.

The 2002 workshop named certain specifics, and Netconf met the goal of
an XML-based configuration protocol.
Maybe it's time for IAB to hold another workshop and get feedback,
especially from operators, about where the IETF seems to be going and
whether operators think we are heading in the right direction by
expanding the nature of IETF OAM and moving significant portions of
the OAM work into other areas of the IETF.

Configuration is obviously a very dfficult problem, partly because we
have a moving target. 
The Internet keeps changing and we need to keep up with the changes
(albeit with some necessary delay).
The IETF already has over 100 RFCs that deal with "configuration".
I think the fact that "OAM" is being addressed in areas other than OPS
reflects that OPS has had a long tendency to have a singular view of
what OPS and management means, and that may not match the operations
perspective of the industry.

Tena Tsou recently published a draft about configuration. I don't
think the draft was very well written, but hidden in the draft was an
important point - many vendors in the telecomm industry are moving
toward an all-IP strategy. Telecom networks, such as LTE, will build
atop the Internet, and need some standardzied, automated configuration
management. 

The discussion about declaring COPS-PR Historic raised an interesting
point. Cable modems and other equipment will build atop the Internet,
and need configuration using some type of automated, policy-based
management. 

Do we really believe that Netconf and SNMP address those configuration
needs adequately? I don't see either being adequate to handle
automated policy-based configuration. Are we even working on that
problem in OPS?

> The lack of any real commitment by the IESG to either NETCONF or
SNMP
> (for this function) is not helping.

I don't think IESG commitment is that important. I think market
commitment is important. I think addressing real market needs is
important. I think we need to start the discussion to understand what
the emerging needs are so we can begin to address them.

> 
> 
> Andy
> 
> 
> 
> > There are a few points I think did not get properly 
> represented in the
> > minutes.
> > The audio archive recorded what was actually said.
> > 
> > Dave H. Why is OPSAREA needed? OAM is done in many several groups
> > forum. We have little real discussions.
> > 
> > more accurately: I think it might help if we tried to drive some
> > discussion in the OPS area. I am beginning to question what the
OPS
> > area is doing in the IETF now. Operators don't come to the IETF
> > meetings very often. As we move to the multi-protocol NM model,
lots
> > of OAM work is being done elsewhere in the IETF. If we don't do
> > operations, and management moves elsewhere, I am not sure 
> what the O&M
> > area is for.
> > 
> > and ...
> > Dave H. There could be issue, OPSWG is where multi-protocol 
> issues are
> > discussed without real plan/framework. We need to decide whether
we
> > want to push YANG. We need to push this stuff. My company is
looking
> > at it. 
> > 
> > more accurately: 
> > Dave H: The OPSAREA session could identify an issue that should be
> > discussed. For example, what does the community want to do 
> about this
> > emerging multi-protocol management approach? We should get 
> some debate
> > going. Right now, there is no discussion going on. 
> > Dan R: Discussion has been limited because we needed to get some
> > pieces into place. Now that we have RFC5706 Guidelines, and YANG
and
> > the YANG architecture moving forward, we are in a better position
to
> > discuss multi-protocol management. This type of discussion is
value
> > that the OPS area can bring to the IETF. The OPSAWG WG has 
> started to
> > work on these directions; we still have work to do.
> > Dave H: I think there are a few important issues we should be
> > discussing, probably in the OPS open area meeting. The OPSAWG WG
is
> > defining how protocols should work together, such as syslog 
> and SNMP.
> > But this work is being done piecemeal without any framework or
real
> > plan about where we are going with this. YANG is about to 
> be submitted
> > for standards track. I think we need a discussion of whether we
want
> > to start encouraging other WGs to develop YANG models or not. I
> > haven't seen that discussion except in the Netconf and Netmod WGs,
> > where there is not surprisingly some bias on that question. 
> We need to
> > have some of those discussions and they are not being driven in
the
> > OPS area. My company is becoming concerned about the (lack of)
> > relevance of the work going on in the OPS area.
> > 
> > David Harrington
> > [email protected]
> > [email protected]
> > [email protected]
> > 
> > 
> > 
> >> -----Original Message-----
> >> From: [email protected] 
> >> [mailto:[email protected]] On Behalf Of Romascanu, 
> Dan (Dan)
> >> Sent: Wednesday, December 02, 2009 8:19 AM
> >> To: OPS Area (E-mail)
> >> Subject: [OPS-AREA] OPSAREA - preliminary minutes of the 
> >> Hiroshima sessionuploaded
> >>
> >> I have uploaded the preliminary minutes of the OPSAREA open
meeting
> > in
> >> Hiroshima. Please read and send me any correction or
supplementary
> >> information that I may have missed. 
> >>
> >> http://www.ietf.org/proceedings/09nov/minutes/opsarea.txt
> >>
> >> Thanks and Regards,
> >>
> >> Dan
> >> _______________________________________________
> >> OPS-AREA mailing list
> >> [email protected]
> >> https://www.ietf.org/mailman/listinfo/ops-area
> >>
> > 
> > _______________________________________________
> > OPS-AREA mailing list
> > [email protected]
> > https://www.ietf.org/mailman/listinfo/ops-area
> > 
> 
>