On Feb 27, 2013, at 11:22 AM, Mattia Rossi <[email protected]> wrote: > I believe this is what Wide Area Bonjour does these days: If you use mDNS on > one Network/Domain, you'll end up asmydevice.networkone.net, and if you move > to another Network/Domain you'll end up as mydevice.networktwo.net
This is certainly less bad than what CeroWRT does with mDNS across multiple subnets within the homenet now. > Still, someone from a third WAB network knows, that it's the same device (if > the device name is unique enough - but in homenet that shouldn't be a > problem). How do they know this? And if they can know this, why not just use the same name in both places? Would it be easy for me to create an attack based on the tendency for people to make assumptions like this, if such a tendency in fact exists? > WAB works using a unicast DNS server, Long Lived Queries and Dynamic Updates > per network segment. I don't think that's the way to go in the homenet > though. WAB is intended to cross the boundaries of the home network, while > we're talking what to do within the boundaries. Yup. > If we could cross subnets without DNS and without confusing domain names it > would be better I think. > But as we might want to allow people who want a domain structure to set it > up, it might be a good idea to recommend the inclusion of a DNS server > capable of dynamic DNS updates in each homenet router... > Can be switched off by default... If it's switched off by default, it won't work, so there's no point in specifying it. I think what CeroWRT (and perhaps OpenWRT) do here is actually exactly right: it's on by default, but just uses a restricted domain. If you want global naming, you can set it up. I think more work on making this easy for a non-geek end-user to set up is worth doing, but I think it's actually a much better basis for naming on a homenet than mdns. > Might be something for a future draft "naming of devices in the homenet". Is > anyone already working on something like that? We have Stuart's draft, which actually makes me even less enthusiastic about mDNS in this context, since apparently Apple's patented the use of mDNS to solve this problem. Lee Howard's written a draft that thoroughly analyzes the state of the art for dealing with the rDNS side of this, and I wrote a draft talking about how to make rDNS work with DHCP+PD. I don't think anyone has talked about how to handle the forward zone, other than what you've seen here from Dave Täht and Simon Kelley, which I think is a very good start and ought to be written up. _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
