On Thu, May 01, 2014 at 10:57:48AM -0400, Jeffrey Haas wrote: > On Thu, May 01, 2014 at 07:38:39AM +0200, Juergen Schoenwaelder wrote: > > On Wed, Apr 30, 2014 at 08:24:51PM -0400, Jeffrey Haas wrote: > > > > > > This is where I think we're getting stuck. There is some vagueness > > > between > > > operational state that can be modified and ephemeral configuration. > > > > Are you saying there is a difference between 'writable operational > > state' and 'ephemeral configuration' but the difference is still > > unclear and hence vague? > > Yes and no. I'm still absorbing appropriate terminology from the N/Y RFCs > so please have patience with me. > > State that I2RS injects into a system may be intended to be ephemeral. > That state may overlap with state that is otherwise configured in the usual > configuration based datastore. E.g. static routes. > > It is clearly possible to generate yang for I2RS that duplicates as much of > the yang for that configured state (e.g. static routes). Yang provides > mechanisms to allow for re-use of the yang inputs (e.g. grouping) in another > module, provided that the original module was authored to enable re-use. > > My observation is that in cases where there may be significant overlap of > I2RS injected ephemeral state, e.g. static routes, that it might be > desirable to not have to have a separate I2RS yang module unless there are > differences that necessitate a new module. Effective, if I have my > terminology correct, to permit more than one datastore to be manipulated > using the same module. I.e. "running configuration" and "I2RS ephemeral". > > If such a thing were possible, then it would only compound the issues noted > in draft-bjorklund-netmod-operational since the configuration state for the > overlapping datastores also had related potentialy overlap in the > operational state. > > To offer an older analogy for comparison, if I can configure static routes > via CLI and simultaneously by SNMP and both configurations can serve as > inputs to the RIB, the installed static route may come from either of the > candidates and the active entry will be selected by a preference. > > In such a scenario, I still need to be able to see my configuration for the > CLI route and the SNMP installed route along with the actual route installed > in the RIB as active.
All fine but the question was whether you see a difference between 'writable operational state' and 'ephemeral configuration'. You are concerned about YANG module organization, how to maintain generic reusable parts and stuff that is config or i2rs specific. And yes, this requires careful engineering and we will not always get it right (and in particular in the IETF this might happen since WGs run at different pace). But on the other hand, if I2RS chooses a different technology, there will likely be no reuse at all (and we could only hope to at least have consistent underlying information models). /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
