Re: [Rtg-yang-coord] Requirements for I2RS protocol and I2RS interim (6/24/2015 at 10:00 - 11:30am ET)

[Jeffrey Haas <[e-mail]>]

Lou,

On Wed, Jul 01, 2015 at 11:38:36AM -0400, Lou Berger wrote:
> > At this time, none of the Topology models utilize Traffic
> engineering.  It is anticipated that these models will support traffic
> engineering.
> 
> Sue, Jeff,
>     Given some recent messages on the list, I thought it best to confirm
> we're all still on the same page. 
> 
> Based on our past discussions, it's my understanding that the work being
> done in TEAS (see draft-ietf-teas-yang-te-topo) will serve as the
> foundation for TE support in the I2RS models.  I complete agree that
> there are details on how the different topology models will
> integrate/mesh still needs to be worked out, and I assumed that this is
> what you/Sue were referring to above. -- But we will not have two sets
> of TE topology models.

I'll admit to not having recently read the teas model.  I'll try to correct
that soon.

That said, we have some interesting headaches to work through for i2rs
purposes:
- We need a base model to work from.  Which group provides the base is
  unimportant.
- Some level of the abstract TE interactions that have i2rs impacts will
  potentially impose some impact on the models.
- The biggest question is interactions with regard to ephemeral state.

As you might have noticed, we're having quite the issue trying to shake out
ephemeral state issues with netmod/netconf.  Until we know what the answer
is going to be in yang and the protocols, there's a large unknown as to
whether we'll need our own "base model" or not.  

But just like with everything else in yang, we want to make sure that when
there are reusable components, lets do it one place.  As I said during
today's i2rs interim, we do have to be careful about doing premature refactoring -
doubly so as we don't know what the impacts of ephemeral stuff will be on
i2rs.

(Benoit will hopefully take the above observation as a note that even
modeling work is starting to get impacted now and not just language/protocol.)

The best I can recommend at this time is lets make sure everyone is paying
attention to both drafts, identifying common components and be aware that we
may need to reconcile the models via refactoring when the time comes.  Plus
it gets more eyeballs on both models.

> I believe that the authors of the respective drafts are already are
> working this.  (Pavan, my co-chair as well as coauthor of the TEAS yang
> model, can comment on this.)

I'm going to largely say, "Don't Panic".  We all want to do the right thing
IETF-wise.

-- Jeff