Re: [Softwires] map-t to proposed standard rather than experimental

[Rémi Després <[e-mail]>]

13 nov. 2014  22:40, Ted Lemon <[e-mail]>  :

> On Nov 13, 2014, at 11:32 AM, Rémi Després <[e-mail]> wrote:
>> (a) Incompatibility with path MTU discovery  (standard-track RFC4821) is, in my understanding, sufficient a reason to keep MAP-T experimental. 
> 
> Pretty much everything on the internet is incompatible with ICMP-based PMTU discovery, unfortunately.

Indeed.
That’s why the packetization-layer PMTUD has been specified (precisely in the RFC4821 I mentioned).
It says (ref. http://tools.ietf.org/html/rfc4821#section-4)
"All hosts SHOULD use IPv4 fragmentation in a mode that mimics IPv6 functionality.  All fragmentation SHOULD be done on the host, and all IPv4 packets, including fragments, SHOULD have the DF bit set such that they will not be fragmented (again) in the network."


>   But can you explain the specific issue here?   Is the problem present only for MAP-T and not MAP-E and lw4over6?

- Only MAP-T lacks transparency to the IPv4 DF bit. 
(Encapsulations of MAP-E and lw4over6 preserve the DF bit; 4rd has a specific provision for DF transparency.)
- This problem has been known since the interim Softwire meeting in Beijing, in September 2011.
- The WG has been made aware at various occasions (see in particular the first bullet of slide 4 in http://www.ietf.org/proceedings/83/slides/slides-83-softwire-12.pdf) 


Regards,
RD