Hi Joe, > -----Original Message----- > From: Joe Touch [mailto:[email protected]] > Sent: Wednesday, June 22, 2016 12:07 PM > To: Templin, Fred L <[email protected]>; Vincent Roca > <[email protected]> > Cc: [email protected]; [email protected] > Subject: Re: [Int-area] Comments for draft-ietf-intarea-tunnels-02 > > > > On 6/20/2016 7:52 AM, Templin, Fred L wrote: > > Hi Joe, > > > >> -----Original Message----- > >> From: Joe Touch [mailto:[email protected]] > >> Sent: Friday, June 17, 2016 2:47 PM > >> To: Templin, Fred L <[email protected]>; Vincent Roca > >> <[email protected]> > >> Cc: [email protected]; [email protected] > >> Subject: Re: [Int-area] Comments for draft-ietf-intarea-tunnels-02 > >> > >> ... > >> (yes, I'd be even happier if all networks periodically tested the > >> minimum MTU as part of operations monitoring) > > The problem is that the ingress is not the source of the original packets; > > it is only the source of the tunneled packets. And, the ingress may need > > to encapsulate packets coming from many original sources that may have > > many different flow label and/or TOS values. Therefore, there is no way > > for the ingress to ensure that any probes it sends will travel over the > > same paths that ordinary data packets will travel due to multipath. > > For the benefit of others, Fred and I have discussed this before.
I guess maybe for the past 15yrs or so? :^} > My position is that a network path is determined by network header > reachability. > > Fred - your point is correct, but it also means that there is absolutely > no packet that can be used a probe for any other packet unless they are > identical in headers, content, and length. That is right. And, if the ingress is asked to encapsulate data packets that have thousands of distinct flow labels there is no one probe packet that can tell the whole story. The only alternative is to treat all data packets as probes and have the network send a PTB message if a probe is lost. Which is exactly RFC1191/RFC1981 which we know have problems. > If your network "path" is > susceptible to that, then maybe it really should behave unexpectedly. It has been suggested to me that some networking gear may inspect not only the outer packet headers but also inner packet headers when making ECMP/LAG decisions - deep packet inspection. That means that even if the ingress used an identical encapsulation header for all data packets the network could still decide to forward them along various paths of the multipath. I don't know how many years it has been that you have been telling me that fragmentation is inevitable - probably almost as long as I have known you. I tried everything under the sun to come up with a way to avoid fragmentation in all cases (see my expired drafts) but I am forced to conclude that you were right. Thanks - Fred [email protected] > Joe _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
