Hi,
I have read both drafts and I can't say either I like either solution too
much.
Some general observations:
- the term "running config" has been in use since the 80s.
Changing it to "intended config" is not helpful.
- YANG and NETCONF/RESTCONF distinguish between config=true and
config=false nodes. There is no reason to alter these protocols wrt/ to
existing operations. Add new operations with new features.
- metadata maintenance is expensive. The opstate-metadata draft seems
very expensive
to implement and not actually required. 1 simple boolean or
enumeration should be good enough
- any solution based on retrieval and compare or push and compare grows
linearly with
the size of the edit. A notification approach offers O(1)
performance, not O(N).
- a data+metadata design allows push or pull in 1 step instead of 2. Data
cannot get out of
sync if it is always retrieved together.
- the enable/disable feature should be separate, based on a simplified
version of Kent's draft
on conditional enablement
- the server has to be aware of intended vs. applied no matter how this
feature is implemented.
Leave most of it as an implementation detail, and focus on the
interaction model for
determining convergence
- a solution to tag YANG config=false data nodes as operational state vs.
statistics is needed.
(YANG Push should make it easy to configure an on-demand subscription for
just operational state
or a periodic for statistics, or different push parameters for each type
in the same subscription.
The client should not have to cherry-pick the opstate vs. statistics in a
subtree
Andy
On Mon, Jul 11, 2016 at 7:49 AM, Juergen Schoenwaelder <
j.schoenwael...@jacobs-university.de> wrote:
> On Mon, Jul 11, 2016 at 03:30:13PM +0100, Robert Wilton wrote:
> >
> > Juergen, I have tried to keep this as an objective comparison between the
> > two drafts. Of course, If I have misrepresented any of the points of
> > draft-schoenw in any way then please correct my comments above.
> >
>
> I may not be clear about everything but I think it also does not help
> if we start a long thread about this. What is more important, I think,
> is that people read both IDs and try to understand the differences as
> they are in the IDs. And then we might have an informed audience in
> Berlin.
>
> /js
>
> --
> Juergen Schoenwaelder Jacobs University Bremen gGmbH
> Phone: +49 421 200 3587 Campus Ring 1 | 28759 Bremen | Germany
> Fax: +49 421 200 3103 <http://www.jacobs-university.de/>
>
> _______________________________________________
> netmod mailing list
> netmod@ietf.org
> https://www.ietf.org/mailman/listinfo/netmod
>
_______________________________________________
netmod mailing list
netmod@ietf.org
https://www.ietf.org/mailman/listinfo/netmod
