Hi, Kent: -----邮件原件----- >发件人: netmod [mailto:[email protected]] 代表 Kent Watsen >发送时间: 2015年6月18日 9:24 >收件人: [email protected] >主题: [netmod] review of draft-openconfig-netmod-model-structure-00 > > > [As an individual contributor] > >In preparing to tomorrow's virtual interim meeting, I reread the >model-structure draft with an eye towards understanding the problem >statements. So far as I can tell, there are four problem >statements: > > 1: disparate models exist without guidance how they stitch > together > > 2: some uber modules, importing many other modules, exist > but are incomplete > > 3: a multi-rooted tree that does not follow any logical > construction is difficult to work with operationally > > > 4: no model catalog/registry is available to discover models > > >My issue with these problem statements is that they are provided without clear >examples illustrating the actual problems. Some references to drafts are >provided, but it's left to the reader to >determine what the problems are. For instance, > > - for (1), RFCs 7223 and 7317 are provided as an example, but > without explanation. Is the issue that interfaces defined > in 7223 would be used by the NTP server defined in 7317? > > - for (2), drafts-ietf-netmod-routing-cfg is provided as an > example. Looking at it, I see that YANG module > ietf-ipv6-unicast-routing imports a number of other modules, > but it's not clear why there is an expectation that it > should be any more than it is. > > - for (3), RFCs 7223 and 7277 are provided as an example, > which I found confusing as 7277 actually augments 7223. > But it may be that this example was meant to highlight > a rare case where a good thing occurred (the text isn't > clear and, if so, a read it wrong the first time)
Interestingly, Section 4.2 (service-layer composition) describes what I view to be the expected solution to many of these issues, and yet it's not described as such. [Qin]: service layer composition follows top down approach, another approach is bottom up approach or map network technology specific model to technology independent model, what model structure draft proposed solution, in my understanding, is pull vs push approaches discussed in the section 3, one example for pull vs push approach, I think is drafts-ietf-netmod-routing-cfg follows push approach while draft-shaikh-idr-bgp-model-02 follows pull approach, I am wondering how to come up hybrid construction which combines both approaches. But it will be interesting to see that happen. This implies that there may be yet another problem statement missing in my list above. [Qin]: Another problem statement is linking to l3sm work and YANG classification work. >Thanks, >Kent _______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
