Re: [manet] NHDP: Link quality based on seq-num

["Teco Boot" <teco at inf-net.nl>]

Hi Ulrich,

>-----Oorspronkelijk bericht-----
>Van: Ulrich Herberg [mailto:ulrich at herberg.name]
>Verzonden: maandag 9 november 2009 23:45
>Aan: Henning Rogge
>CC: manet at ietf.org; Teco Boot; Dearlove, Christopher (UK)
>Onderwerp: Re: [manet] NHDP: Link quality based on seq-num
>
>On Mon, Nov 9, 2009 at 11:37 PM, Henning Rogge
><henning.rogge at fkie.fraunhofer.de> wrote:
>> On Mon November 9 2009 15:31:16 Teco Boot wrote:
>>> What if someone defines a reliable transport and suppress
>>> repeating messages?
>>> Are we sure we want to use missing sequence numbers as
>>> indication of missed packets?
>>> What if other reasons caused the loss, e.g. tail drop during
>>> congestion caused by the MANET protocol itself?
>
>> Packet Sequence numbers have the same "range" as a linklocal IP, so I
>don't
>> see any problem with reliable transports. Such a transport would be
>built
>> "below" packetbb, so the mechanism should be transparent to packetbb.
>
>I don't see the relation to LL here. I think what Teco means is that
>if NHDP runs on a reliable transport, a very bad link (that needs many
>retransmissions of HELLOs) still would appear as perfect link to NDHP,
>because no packetbb packet is lost. So, of course the transport
>protocol is transparent to NHDP, but using packet seq. no. as measure
>of number of lost packets would not make sense any more.

Yes, this is an example the mechanism breaks.
I had IP layer in mind, like PIM over TCP. PIM started with repeating
advertisements, but for scalability and other reasons, the protocol is
extended for using reliable transfer (draft-ietf-pim-port).

In my MANETs, I have low bandwidth high costs links and cannot afford 
the repeating MANET messages over it. With reliable multicast (or 
unicast, if the link is P2P), I could suppress. Sequence numbers may 
break this idea.

Teco.  


>
>Ulrich