Re: [NSIS] Approved: draft-ietf-nsis-ntlp-20

[Adrian Farrel <Adrian.Farrel at huawei.com>]

Hi,

With the approval of draft-ietf-nsis-ntlp-20 as an Experimental I-D, I 
wanted to give my own personal view of why this is a good thing, how we 
might have made progress that the working group might have found more 
acceptable, and what could happen going forward.

Firstly, it is important to understand that an Experimantal status is not 
perjorative. In fact, many good and widely deployed protocols grew up as 
Experimental, benefitted from cautious deployment and feedback, and were 
taken forward on the Standards Track. In some cases the protocols went 
forward largely unchanged and in some cases significant updates were
made.

I believe it is really important that GIST will now be published as an RFC. 
This will open up the implementation base and allow people to understand 
that they are implementing a stable specification. In as much as the choice 
of Experimental Status has enabled the log-jam to be broken and publication 
achieved, it cannot be anything except a good thing.

My own view is that we could have reached this stage at least two years ago. 
I am not interested in the who or why we did not, but IMHO we could have 
reached the approved status by either:
- going straight to Experimental status
- clarifying the intended deployment scope.

The latter case needs some clarification. We hear talk of "walled gardens" 
and what is meant is that some protocols are not necessarily suitable (or 
intended) for deployment in every conceivable network. Of particular 
interest are the distinctions between all-new nodes and new/legacy mixes, 
enterprise and service provider networks, sinle domain and multi-domain 
networks, and contained networks and "the Internet". I believe that had we 
made these statements of intent and scope in the GIST spec up front and in 
14pt type, we would have made far better progress.

Of course, the value of this information is now moot. I am not sure it is 
worth discussion as it is picking over old bones. But maybe we can learn for 
next time.

The NSIS working group already has some work on applicability. I think it 
would be really helpful to pursue that thread with further applicability 
stetements that caveat the environments in which GIST should be applied and 
those in which it is not intended to be applied. (This is clearly a choice 
for the WG and the responsible ADs, and not for me!)

I also hope that the folks who have built and deployed GIST will come back 
(possibly in a relatively short period) to discuss their deployment 
experience. Hopefully they will be able to tell us about the extent of their 
deployments in live networks and report on the issues and successes they 
had. The combination of these discussions and the applicability statements 
could possibly be a revision of the GIST RFC that includes statements about 
any deployment limits or advice, any protocol fixes and clarification, and 
movement onto the standards track.

Cheers,
Adrian

PS Caveat! I am just one AD. This is my apple-pie view of the world.