Hi Emmanuel,

From: [email protected] [mailto:[email protected]] On 
Behalf Of Emmanuel Baccelli
Sent: Thursday, January 22, 2015 1:54 PM
To: Templin, Fred L; Internet Area
Cc: Joe Touch; Carsten Bormann
Subject: RE: [Int-area] About draft-baccelli-manet-multihop-communication-04


Hi Fred

Le 22 janv. 2015 19:36, "Templin, Fred L" 
<[email protected]<mailto:[email protected]>> a écrit :
>
> Hi ,
>
> IP runs over "links" as specified in "IP-over-(foo)" documents. What the 
> document in question
> is describing is a "multi-link" which is not a candidate for "IP-over-(foo)".
>

Nevertheless, people (MANET, ROLL, AUTOCONF, 6LOWPAN...) have been running IP 
over interfaces with connectivity as described in this draft. It's difficult 
and it yields specific protocol design decisions because of the unique 
characteristics of such connectivity, but it's doable.

I know a little bit about MANET and AUTOCONF. My understanding was that 
standard routing protocols like OLSR and OSPF-MANET are used to connect these 
kinds of links into multilink subnetworks (note I did NOT say “subnet”) and 
that IP can be run  by configuring an NBMA tunnel that sees the subnetwork as a 
fully connected link. IP then runs in the normal fashion over the NBMA tunnel. 
You can write an “IP-over-(foo)” document about a tunnel of that nature. But, 
this document cannot present itself as an “IP-over-(foo)” document.

Thanks - Fred

More people (6LO, CORE, LWIG...) want to design more protocols running over 
such networks. There is a risk that protocol design decisions that are made do 
not capitalize on available knowledge about the characteristics of such 
networks. Actually, the fact that we are having this argument right now shows 
that:

- there is no such document available yet (except this draft)
- there is a need for such a document

Best,

Emmanuel

> Thanks - Fred
>[email protected]<mailto:[email protected]>
_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to