In message <20140304234206.gc9...@mx1.yitter.info>, Andrew Sullivan writes: > On Tue, Mar 04, 2014 at 04:50:33AM -0800, SM wrote: > > correctly the answer in the DNS cache would be for a non-global > > resource. There is no longer an assumption of uniqueness. As a > > local decision I would use low TTLs [1]. > > But as Michael pointed out upthread, the more realistic and compelling > (if harder to understand) examples involve a dual-homed node like a > phone flipping from one interface to another. You'd need sub-second > TTLs for that not to be a problem, and we don't have those.
I don't see current phones flip from WiFi to 3/4G multiple times a second. They have longer duty cycles than that. Additionally if you are in such a zone you tend to 1) move or 2) force WiFi or 3/4G to get stability. Additionally the impact of getting the wrong version is that you end up with a unusable address when you get a ULA/RFC 1918 address packets to which should be dropped by the telco or you get NXDOMAIN or the global address of the resourse and hair pinnig in the NAT helps with the later. Mark > Best, > > A > > -- > Andrew Sullivan > a...@anvilwalrusden.com > > _______________________________________________ > homenet mailing list > homenet@ietf.org > https://www.ietf.org/mailman/listinfo/homenet -- Mark Andrews, ISC 1 Seymour St., Dundas Valley, NSW 2117, Australia PHONE: +61 2 9871 4742 INTERNET: ma...@isc.org _______________________________________________ homenet mailing list homenet@ietf.org https://www.ietf.org/mailman/listinfo/homenet