Dean,
See in-line.
Benoit,
You describe the situation really well, but as next step we have to get more organized and more coordination in modeling work. We have to make sure that there is not too much overlap between efforts, not too waste resources ... and inconsistencies between data models, which is my primary concern.Let's take one example: policies. I see policies in many YANG models. How do we make sure the policie definitions are consistent and extensible? How can we coordinate this (huge) effort? Who is in charge? The collective responsibility might not be enough…
Coordination is only one issue: its operational efficacy that is also of paramount importance here. We should do what we’ve done in NETMOD's “design teams”: pair up people that are truly implementing functionality in real or virtual devices with operators to ensure that the models created are grounded in the realities of implementation and deployment.
—Tom
and as well create an architecture description document, where the relationship between models will be described. With that we can see which models can be reused as base model in other models. Good example of coordination is the rtg-coord-yang mailing list, but we need one for all IETF, to be able to find what are other the teams working on.
Right.Regards, Benoit
We should not let this momentum go away and with few simple steps (like improving coordination and system meta architecture), continue to feed positive energy into this situation.
|