Re: [v6ops] New draft at dnsop a bis for DNS IPv6 Transport Operational Guidelines

[Philip Homburg <[e-mail]>]

>If I understand correctly, the gist of the argument is that too
>many IPv6 resolvers run with pre-DNS-flag-day-2020 values for
>EDNS0 buffer size (typically 4K), and too many of these actually
>fail to receive fragmented IPv6/UDP responses, and the argument
>is that this together should hold us back from issuing a blanket
>recommendation to use IPv6 for DNS transport at the publishing
>side.

What seems to be mostly ignored in this discussion is that an authoritative
DNS server is free to truncate (and set the TC flag) a UDP response at 
something smaller than the buffer size advertised by the resolver.

So even if the resolver advertises 4KB, the auth. server can still truncate
at something less then 1500 octets.

So to properly evaluate the effect of fragmentation issues, we need to look
not only at resolvers but also at the behavior of authoritative servers.

Where a recursive resolver can be fire and forget, an authoritative server
tends to be essential to business.