On 24/06/2013, at 8:41 AM, Fred Baker (fred) wrote:
>
> I'm in a similar case with respect to protocols above IPv6 (OSPF and NFS/UDP
> come quickly to mind) that depend on fragmentation to deal with the issue. I
> think the Robustness Principle tells us that such applications SHOULD figure
> out how to live with PMTU, but it also tells us that we can't deprecate
> fragmentation unless all known instances that depend on it have defined
> practical work-arounds. I suspect that this would imply the re-creation of
> the fragmentation feature in an intermediate protocol, which seems like a lot
> of work with little real gain.
Calculating the PMTU for UDP protocols is left to applications, the sockets API
doesn't do this under the hood like it does for TCP. As a result most UDP-using
applications don't probe the PMTU. This is the case for both IPv4 and IPv6,
although the effect with IPv6 is lost connectivity rather than in-flight
fragmentation.
Some small changes to the socket API for unconnected protocols would make life
substantially better for RADIUS implementors (and, I imagine, DNS). We're more
than happy to repeatedly pass the same application protocol data unit into the
socket whilst the socket probes the path MTU.
-glen
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
