Source VRF or leafref to ietf-network-instance

[ericvoit]

We have a choice or source-vrf as

(a) a string or (b) a leafref to ietf-network-instance model.

Nothing is wrong with ietf-network-instance model, but it adds the following:

• timeframe for WGLC
• draft has a dependency on schema-mount
• no insights into industry adoption

For the short term at least, (b) adds extra complexity, as the subscribed-notifications model will have to include custom augmentations for vendors to support different VRFs.

[ericvoit]

Based on the discussions completing, I believe there to be rough consensus on the configuration of a VRF for a configured subscription being via a leafref to network-instances with vrf-support being an if-feature. I.e.:
+--ro source-vrf? -> /ni:network-instances/network-instance/name {supports-vrf}?

Based on the planned time frame for network-instances.yang going to the IESG, I don't expect there to be any timing issues. But if it does turn out that there is a delay in this draft, this issue many need to be revisited.