Ron asserts the following two statements in this 00 draft: "..." As a result, IPv6 fragments carrying TCP payload are rarely observed on the Internet.
and "..." Because many UDP-based applications follow the above-quoted recommendation, IPv6 fragments carrying UDP traffic are also rarely observed on the Internet. I'd like to understand the basis of these assertions. I believe what I am seeing, on the edge, suggests there is in fact V6 fragmentation in both TCP and UDP. Ron also asserts: The following is a list of UDP-based applications that do not follow the recommendation of [RFC5405] and rely in IPv6 fragmentation: o DNSSEC [RFC4035] The effectiveness of these protocols may currently be degraded by operator behavior. SeeSection 2.4 for details. I'd like to point out that at least BIND imposes a 1280 IPv6 packet limit. Its wired into the code. -George
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------