Remi, > -----Original Message----- > From: [email protected] [mailto:[email protected]] On > Behalf Of Rémi Després > Sent: Thursday, March 04, 2010 5:26 AM > To: Brian E Carpenter > Cc: [email protected]; Templin, Fred L > Subject: Re: [Softwires] SOFTWIRE working group last call on 6rd > > > Le 4 mars 2010 à 03:23, Brian E Carpenter a écrit : > > > On 2010-03-04 15:13, WashamFan wrote: > > > >> Why advertise WAN side MTU to LAN side? Doesn't routers > >> advertise LAN MTU via RA? PMTU would work out the optimal > >> MTU for end-to-end communication. > > > > Because PMTUD is not reliable. > > The dilemma seems to be: > - either one relies on ICMPv6 packet-too-big error messages being properly > forwarded and processed, > - or, for IPv6 to work, hosts have to be clamped to 1280. > > This is IMHO a good reason to insist, again and again, that ICMPv6 PTB > messages MUST be always > forwarded in intermediated nodes, and always processed in their destination > hosts for PMTUD to become > reliable.
I agree that customer end systems and middleboxes within the customer network *must* be made capable of honoring PMTUD; otherwise, there is no hope. The difficulty lies in reliance on PMTUD for detecting MTU restrictions that occur *outside* of the site. That is where Brian is correct in pointing out that PMTUD is unreliable. That said, it would be very useful to reduce the incidence of invoking PMTUD within the end site if possible without the side effect of clamping MTUs that don't need to be clamped. That is an area in which SEAL can help. Thanks - Fred [email protected] > RD > > _______________________________________________ > Softwires mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/softwires _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
