On Tue, Aug 02, 2005 at 06:49:50PM +0100, Tim Chown wrote: > Hi, > > I raised this question today but we were short of time. > > My concern is that in an enterprise deployment, I might want to avoid the > complexity of privacy addresses (from the management perspective).
Right, I'm not sure if that necessary relates to source address selection though. In such cases I think it better that the interfaces get no privacy addresses at all. [...] > I appreciate Keith's point that per-application (non) usage may also be > desirable, but there is an API being proposed for that? It should probably > have some relationship to the site policy though? In particular for privacy addresses I see a need for per-application behaviour. One concern I have is how applications can distinguish privacy addresses from others. There are apps that need to e.g. find all addresses on an interface or host, and then register some of those somewhere (or pass them as a referral or otherwise to a peer/server). It probably would like to use a stable address and not privacy address for this, but how does it find which are which? Apps shouldn't need to try to guess from the interface id... In the above setting I also see issues with "normal" global vs ULA. An application may need to distinguish them and somehow know which to use. Distinguishing them is easy, knowing which of them to pass may not be trivial. Stig -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
