HI Joe, > -----Original Message----- > From: Joe Touch [mailto:[email protected]] > Sent: Friday, April 24, 2015 10:14 AM > To: Templin, Fred L > Cc: Ronald Bonica; [email protected] > Subject: Re: [Int-area] I-D Action: draft-ietf-intarea-gre-ipv6-07.txt > > > > On 4/24/2015 9:39 AM, Templin, Fred L wrote: > ... > > BTW, the AERO spec already does this kind of probing. > > And yet we are *still* not talking about AERO.
I have pointed you to AERO repeatedly and you seem to be reaching some of the same conclusions that are already in AERO. You are saying use a positive confirmation of a successful probe. AERO has been saying that for a long time now. You are saying do something drastic if probes fail - AERO says fragment and you say shut down. But, both are just drastic measures that are taken only if probes fail. >From this discussion, I have learned that AERO should really be doing RFC6438-style flow label assignments, meaning that ECMP/LAG can take the packets across different paths - all of which need to be probed. But, that is really just an augmentation of what is already in AERO. I have also learned that probing for 1280 is sufficient in environments where it is operationally reasonable to expect that PMTUD will work, whereas probing for 1500 is needed when there is no such operational assurance. > Can we please get back to the subject of this doc? I have indulged your line of discussion in significant detail. You owe me a critique of AERO. Thanks - Fred [email protected] _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
