Hi, a few days ago I noticed on my NTP servers that answers to IPv6 NTP queries are rejected by "stupid" firewalls between the internet and the client. Sometimes I see up to 5-6 % ICMPv6 traffic saying "unreachable port" or "prohibited", mostly coming from 2a02:c78::/29 (Sky Broadband, UK) and 2800:370::/32 (CORPORACION NACIONAL DE TELECOMUNICACIONES, EC).
Maybe the most problems people experience with NTP over IPv6 are related to improper (stateless?) working firewalls which are rejecting or dropping answer packets. The exact same thing is also happening on my authorative DNS servers - also with DNS requests originating at university networks... The overall IPv6 experience regarding latency and stability should increase since you have no NAT layer between the client and NTP server. In Germany, a growing number of users of internet via TV cable are connected via "DS lite" - with IPv4 via carrier NAT and native IPv6. On those connections you have a big difference in latency between IPv4 and IPv6 since IPv4 needs to be translated on an overloaded NAT router. Greetings Max Am 29.06.2016 um 21:20 schrieb Franck Martin: > Really? http://ipv6bingo.com/ > https://blogs.akamai.com/2016/06/preparing-for-ipv6-only-mobile-networks-why-and-how.html > > The IPv6 packet header is fixed size, does not have checksum, the routers > do not fragment and ECN is more deployed on IPv6 than IPv4 and there is not > NAT,.. > > All that contributes to faster routing processing with less errors. > > Toute connaissance est une réponse à une question. > > On Jun 29, 2016, at 10:39, Charles Swiger <[email protected]> wrote: > > Hi, Dan-- > > On Jun 28, 2016, at 8:52 PM, Dan Geist <[email protected]> wrote: > > Chuck, I don't disagree that overall experience may be better for people > with networks that don't have robust IPv6 capability, but there's nothing > magical about v4 (or v6 either) that makes it "perform better" at layer 3. > > > If you'd like to consider things at layer 3, note that IPv4 normally has a > 20-byte header size, and IPv6 has a 40-byte header. For large packets, the > difference in protocol overhead is not very significant-- about 1%-- but > for a 56-byte NTPv4 packet, using IPv6 means sending about 125% as many > bits over the wire as sending the same payload via IPv4. > > If other factors are held equal, IPv4 is always going to perform better > than IPv6 for NTP because smaller packets mean shorter transmit/receive > times and thus reduced latency for NTP polls. > > (There's nothing magical about protocol overhead, except perhaps pretending > that there isn't a difference. :-) > > Regards, >
signature.asc
Description: OpenPGP digital signature
_______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
