Hi Ian, Thanks for the response, please see my comments inline.
BR, Linhui 2018-01-12 17:17 GMT+08:00 Ian Farrer <[email protected]>: > Hi Linhui, > > I agree with your comments on 1,3 and 5. Comments on 2&4 below. > > Thanks, > Ian > > > > > 2) The BR module should also augments ietf-interfaces since they are > also a tunnel endpoint and it is better to be consistent with the CE model. > And in this way, we do not need to declare the redundant br-instance id or > name anymore. > > [if - From my personal perspective, I disagree. In both of the lw4o6 > implementations I’ve worked with, the v6 address used as the BR’s tunnel > endpoint address is not configured on any interface - i.e. it is used > purely as a destination address for routing traffic to, validating received > encapsulated traffic’s source address and as the source address for sent > encapsulation. > > If an implementation did rely on having an address assigned to an > interface, then this could still be done using ietf-interfaces/ietf-ip in > the normal way and the softwire-br model would be configured with the > appropriate matching address. > [LH]: Well, if there is no such requirements from other vendors, just ignore this comment. And a further question here, is the concept of 'instance' in BR module equivalent to the term of interface? > > > > > > 4) The usage of choice statement is not very clear, why do we need to > use the 'case' and 'feature' statements together? It seems that we only > need one of them. > > [if - Well, a CE may implement binding and/or algorithm. If it implements > one, then there is no choice to be made. If it implements both, then it can > only configure one type hence the choice. If it needed to implement two > types for whatever reason, then there would be two instances of the model > attached to different interfaces. > [LH]: That sounds strange, why not just restricting the CE/BR instance (not the actual device) to be only one of them (i.e. binding & algorithm)? In this way, only 'if-feature' is needed.
_______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
