Hi Juergen, Martin,

thank you for your comments!  Couple of brief responses.

-re: intended status experimental.  This is clearly a mistake/oversight.  Will 
fix this.

- re: dividing the model into two modules.  I agree it is possible to put them 
into one module - this is what we had in an earlier revision in fact together.  
The reason why they were separated was to allow for a common umbrella for 
network inventory and topology - you could use the network module also for a 
network inventory (contained in the nodes), without additional topological 
information.  This is how Open Daylight chose to use it.  I share your 
sentiment that in general both will be implemented together, and you could 
still integrate inventory with a combined module (even without populating the 
topology portion - links, TPs - if so desired).  Ultimately, both are viable 
options; it's a design decision and it would be interesting to hear more 
feedback on this; as mentioned there is implementation precedent in ODL.  

- re: the identifier/referencing scheme, and potential use of 
instance-identifer as alternative.  The requirement is to express the 
constraint that  the references are to instances in the proper "underlay",  
e.g., a supporting-node of a node needs to be in a supporting-network, which 
was reason for navigate the paths. 

--- Alex

-----Original Message-----
From: Juergen Schoenwaelder [mailto:[email protected]] 
Sent: Thursday, April 16, 2015 4:38 AM
To: Martin Bjorklund
Cc: Alexander Clemm (alex); [email protected]
Subject: Re: [i2rs] I-D Action: draft-ietf-i2rs-yang-network-topo-00.txt

On Thu, Apr 16, 2015 at 09:32:48AM +0200, Martin Bjorklund wrote:
> 
> No the argument is not that operators can't understand instance 
> identifiers.  The argument is that leafrefs are more precise and might 
> be easier to read and write than i-is.
>

Yes, instance-identifiers do not support a path restriction (I do not recall 
why we did not consider to have this in YANG - but that does not matter here).

I personally find a single 'pointer' value way easier to deal with than a 
triple of values that together act as a single 'pointer' value.

/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/>

_______________________________________________
i2rs mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2rs

Reply via email to