>>> This obviously only works for systems with coherent forward/reverse >>> DNS information and which use the same name for IPv4 and IPv6. >>> Notably this excludes many DSL/Cable/Dial-up users. (Potentially a >>> large portion of European IPv6 capable hosts.) >>> >> My machine was using different names for IPv4 and IPv6, but I've now >> fixed that so it should now appear in your search. >> (tardis.devnet-uk.net) > > Is that how we'd want it? I wasn't advocating either way. This is just one manner of surveying the IPv6 capability of the pool. It's not supposed to be authoritative, complete, or used as a recommendation for any way of doing anything.
That said... > I remember there were some concerns about DNS queries returning (or > clients preferring) AAAA over A records. DNS servers simply reply with the requested records, it's up to your client to decide what to do with it. Vista prefers AAAA records when it's configured with an actual IPv6 address (i.e. not Toredo or local-link). From what I've observered Linux and KAME-based systems perform the same, but you'd need to look it up. Systems without IPv6 addresses shouldn't be requesting AAAA records anyway. This may or may not be an issue depending on how you see things. Google's survey indicated only 0.09% of users have broken IPv6 connectivity. Having both AAAA and A records for the same name should result no issue for 99.91% of users. (They'll either continue using IPv4 or correctly use IPv6. The remaining 0.09% may be unable to connect or experience delays; I could not find Google's definition of 'broken'.) It's worth noting Google's survey corresponds to a very wide and diverse user base. A proper survey of the pool's client base would likely be needed to determine the client bases IPv6 capability. For the pool's purposes, it would be prudent to study the IPv6 vs IPv4 latency of it's clients as well. While native and tunnel-brokered IPv6 connections are often/usually in the range of native IPv4 latencies, 6to4 may create a ridiculously long rtt as many North American 6to4 anycast tunnels terminate in Sweden. I'm not entirely sure how you'd accurately collect data about the pool's clients, but I would be curious to know how many AAAA queries the pool DNS servers are receiving. > I use different names for v4 and v6 services too, and that seems like > a viable way of creating an "opt-in" system for both users and > servers. Obviously the pool is going to need to pick one system or the other. I'm really only curious how many hosts a search like this turns up, nothing more. Incidentally, currently 53 unique IPv6 addresses, obviously this number should continue rising until the whole pool has been queries (assuming an even distribution throughout the pool and not a higher concentration in host that appear more frequently in the DNS response). Cheers, Kevin Bralten _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
