Sue, On Tue, Sep 23, 2014 at 06:14:16PM -0400, Susan Hares wrote: > The fourth option is interesting and creative. A few questions/requests: 1) > How would the fourth model be expressed in the yang model - config > (empheral)?
I believe Juergen has answered this point. > 2) Please define your definition of local config with an > example., This corresponds to "config true" state, probably in the running datastore. The semantics on the impact for the commit model when writable-running is not in use haven't been fully explored. > and 3) please give an example of the (must, when) use cases > envisioned? In the original proposal in my draft, when we had completely disjoint datastores, consider the BGP I2RS ephemeral peer case. The I2RS model would have a container of ephemeral peers. That container might not specify the local Autonomous System number. In that case, there would be a MUST requirement for that neighbor that there exist a system-wide configured Autonomous System number, probably in the local Config. Currently, there is no syntax that says how to examine nodes that are in a different datastore. From an XPath context, the current datastore is the document context. In the new proposal, since the ephemeral datastore gets a shadow of the local config, there is no need to have extensions to XPath to refer between data stores. However, if we end up with more than one ephemeral datastore (not recommended for this reason), such an issue returns. -- Jeff _______________________________________________ i2rs mailing list [email protected] https://www.ietf.org/mailman/listinfo/i2rs
