Hello Vicky,
I am O.K. with what you said, but I want to reiterate that the *reason*
that RFC 6621 was published was *exactly* so that protocols like AODVv2
could use it. I know this because I was one of the main contributors --
specifically the parts about hashing for duplicate detection and much of
the part about CDSs. Somehow RFC 5444 seems to have gotten in the way.
I also did a thorough analysis of the chances of failure due to hash
collisions and the recommendation makes the failure mode far less common
than packet drops due to interference.
Anyway, the suggestion that RFC 6621 is inapplicable relies on a
distortion of reality.
Regards,
Charlie P.
On 3/3/2016 1:21 AM, Victoria Mercieca wrote:
I believe that we caused a lot of confusion with our attempt to explain this at the time, which is why it was removed from the core draft. Yes, 6621 is inapplicable in that (I think due to the rules John pointed out in another email I just read) the packets would not be deemed to be forwardable by 6621. In any case we dont want them to be forwarded by some type of 6621 implementation since we want AODVv2 to process and regenerate them instead. 6621 also discusses duplicate suppression, which is handled by AODVv2 itself already, plus I'm not sure if the way 6621 determines duplicity would achieve AODVv2's idea of redundancy. Lets not argue about that anyway...because ...
I think what Charlie was getting at was that if AODVv2 knew whether the router was elected as a forwarding router then it could decide whether it should regenerate messages, as a general rule rather than on a packet by packet basis. I think this technique is still covered by statements like "The router MAY choose not to regenerate the RREQ, for example if the router is heavily loaded or low on energy and therefore unwilling to advertise routing capability for more traffic." This covers implementers who wish to be clever with any 6621-inspired decisions, without getting into the details and confusing people.
Charlie's draft explains this, but it seems unnecessary to have a separate draft for such a short explanation. Why don't we explain this to MANET properly, get some feedback, and if they still hate it, leave the draft without any mention, because the statement I quoted is enough (in my opinion anyway) to allow an implementation to use some 6621-influenced technique to decide whether to regenerate.
Kind regards,
Vicky.
On Wed, Mar 2, 2016 at 5:03 PM, Charlie Perkins <charles.perkins@xxxxxxxxxxxxx <mailto:charles.perkins@xxxxxxxxxxxxx>> wrote:
Hello again folks,
It seems like I recall seeing a complaint by Christopher Dearlove
to the effect that RFC 6621 is simply inapplicable to AODVv2. Is
that comment something I should respond to on the mailing list?
I am re-sending the attached draft to show just how wrong his
complaint is. Please let me know if you have any comments.
I sent it out earlier (on November 26 of last year) but did not
get any comments. It's very short. Please take a look.
Regards,
Charlie P.
