In message <[email protected]>, Mikael Abrah amsson writes: > On Tue, 3 Mar 2015, Mark Andrews wrote: > > > What we really should be telling ISPs is that renumber events should be > > make before break. There is zero reason other plain poor customer > > service to not do this. > > There are some markets in the world, where customers *demand* to be > frequently renumbered, because they think this is a privacy measure.
And their long running TCP sessions break in IPv4. This doesn't mean that they can't do better with IPv6. We could also extend the temporary / permanent address model to PD. You get two prefixes from the ISP. One is designed to servers and other things that need long term stability. One is designed for ephemeral use. The application chooses which to use like they do today. > My current thinking in this area is to provide about 1 hour of address > overlap to the home, where the old prefix is not preferred and the new is > available for new connections. This would mean that most services would > work just fine as long as they frequently (at least once per hour) > re-establish their TCP session. Otherwise they have to do like mosh and > try other connections when the first one breaks. > > I still think there needs to be quite a lot of work done on APIs and best > common practices in order for applications to do the right thing so this > kind of renumbering event works. Most likely it's going to require a FOSS > library that will act as a middle layer between the application and the > network so applications don't have to talk directly to the POSIX > interfaces (which plain suck and haven't been updated in 15 years or so). > > We probably need a 5-10 (wo)man-year effort with a FOSS code outcome in > order to get this done. I don't know who's interested in doing this. We > have quite a lot of standards and documentation that can be used, what's > lacking is actual code that the normal application developer can use, that > is also multi-platform. Right now Microsoft, Apple and Google are putting > a lot of effort into these APIs for Windows, OSX/iOS and Android > respectively. We need this for the mainly Linux-driven FOSS universe (I > don't know how much of the Android efforts are actually trickling back > into regular Linux). > > I belive MP-TCP has a role to play here as well. > > -- > Mikael Abrahamsson email: [email protected] -- Mark Andrews, ISC 1 Seymour St., Dundas Valley, NSW 2117, Australia PHONE: +61 2 9871 4742 INTERNET: [email protected] _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
