----- Original Message -----
From: "Pete Resnick" <presnick(_at_)qti(_dot_)qualcomm(_dot_)com>
To: "John C Klensin" <john-ietf(_at_)jck(_dot_)com>
Sent: Friday, March 06, 2015 5:41 PM
Closing the loop:
<snip>
There is a much longer discussion to be had on this topic than I'm
willing to do here, but summarized: This experience puts an
exclamation
point on the thought that I've had for some time, that AD-sponsored
documents, and doing things outside of the context of a WG generally,
is
a really bad idea if you want something standardized in the IETF. The
reason that people have been doing so (whether consciously or just
influenced to do so) is that the WG process and coming to consensus
seemed (and truly was) hugely daunting. But we end up in situations
like
this, where things get developed without the iterative
consensus-building that we're supposed to be doing. The better thing
to
do in order to not discourage others and to create good work is to
improve the speed at which WGs get things done and make it less
daunting, and I believe that we've been doing that of late: We've had
several good examples of WGs that we've spun on quickly to do a
focused
piece of work, and they're done in months, not years. But more beer
required to fully have that discussion.
Pete
I agree and find it timely. I have been noting the tide of I-D with
YANG models being posted and particularly those with names such as
draft-johndoe-netmod-routing-apps-etc
The format of the I-D's name implies, to most of us, that this I-D is
headed for the netmod WG which I think is unlikely to produce a
satisfactory RFC on the topic. Whatever the technology to be managed
is, I think that the prime requirement is to have those expert in the
technology work on it first, producing something at the level of an
Information Model. Get that wrong and you may have a perfect, but
useless, management module, regardless of whether it is in YANG, SNMP,
etc.
I note that one such I-D started out
draft-zhdankin-netmod-bgp-cfg-01
and became
draft-zhdankin-idr-bgp-cfg-01
Spot on, but it could have been right from the word go.
Fine when there is an obvious WG, but what to do when there is not?
Where should
draft-asechoud-netmod-diffserv-model-01.txt
go? I suspect that the netmod WG has few diffserv skills, but where in
the IETF are there any? Not enough to form a WG I suspect. This is the
sort of thing that becomes AD-sponsored but I think that it should not.
Informational? Or should the IETF turn away such work? I don't know.
Tom Petch
pr
--
Pete Resnick<http://www.qualcomm.com/~presnick/>
Qualcomm Technologies, Inc. - +1 (858)651-4478