On 18/11/2016 14:53, Michael Behringer (mbehring) wrote: > One question that just came up: Should Intent be designed per ASA or per AF? > > My suggestion previously was to segment Intent into sections per Autonomic > Functions. > > Example: Intent for the bootstrap function could be: > - allow bootstrapping new devices only during maintenance window > > For such Intent, action could be taken on the registrar (one ASA of the AF), > or on the proxy (another ASA of the same AF). > > It seems to me an author of an AF might like all ASAs of his AF to know about > the Intent, because the proxy may also take actions.
Yes, I think so. Again we hit the question of granularity and whether this is actually Intent, but a policy like that might be added to Intent-for-BRSKI, and both registrars and proxies need to know it. So I'd expect we will need intent to be in sections that are per-function (and we will therefore need a method of guaranteeing that function names are unique). In fact we get a more general result if we simply say that intent is divided in sections that have unique labels; then it's easy to support intent per-function, per-objective, or even per-geography if we want. Brian > > Michael > > _______________________________________________ > Anima mailing list > Anima@ietf.org > https://www.ietf.org/mailman/listinfo/anima > _______________________________________________ Anima mailing list Anima@ietf.org https://www.ietf.org/mailman/listinfo/anima