Re: [Ospf-manet] Re: Ospf-manet Digest, Vol 11, Issue 5

[Richard Ogier <[e-mail]>]

Aniket,

You described very well what I mean by "natural extension".
You are right that it has a lot to do with scalability.
MDRs achieve scalability in an MANET in the same way that DRs
achieve scalability in an OSPF broadcast network.
It would be great if more people would take the time to understand
the MDR approach as well as you do.

Here is a related paragraph from the introduction of the MDR draft:

   OSPF-MDR was motivated by the desire to extend OSPF to support
   MANETs, while keeping the same design philosophy as OSPF and using
   techniques that are similar to those of OSPF.  For example, OSPF
   reduces overhead in a broadcast network by electing a Designated
   Router (DR) and Backup DR, and by having two neighboring routers form
   an adjacency only if one of them is the DR or Backup DR.  This idea
   can be generalized to a multihop wireless network by forming a
   spanning tree, with the edges of the tree being the adjacencies and
   the interior (non-leaf) nodes of the tree being the generalized DRs,
   called MANET Designated Routers (MDRs).

Richard


Aniket Desai wrote:

> At 06:41 PM 10/2/2006, you wrote:
>
>> IMHO, it doesn't matter what you think is a "natural" and "unnatural"
>> extension to OSPF. What should matter is the requirements, and the way
>> to solve those requirements.
>
>
> - Let's say OSPF-MDR is a natural way to extend OSPF's broadcast 
> interface for MANET type physical medium. It should be an acceptable 
> term. What requirements does this solve: scalability, scalability and 
> scalability.
>
>
>> I didn't ask you to stop saying "this is natural" because I have any
>> thoughts on what's natural or not, but rather, because it doesn't lead
>> to any useful discussion. Tossing words with emotional connotations into
>> the conversation doesn't clarify it in any way, shape, or form.
>>
>> I could, in fact, argue that just about _any_ extension to OSPF is
>> "natural," in some sense or another. It's a meaningless concept, in this
>> world.
>
>
> If you agree that OSPF is about maintaining a connected/biconnected 
> adjacency graph that is rooted towards its leaders (DR and BDR) which 
> are also the *forwarders* of data on this graph, then the CDS concept 
> and the subsequent MDRs make perfect sense. That is how scalability 
> should be achieved in OSPF. In this design, each node tries to form 
> adjacencies that are aligned towards the leaders, and the leader 
> *backbone* (i.e. CDS) will forward the LSAs originated by everyone 
> everywhere. Also *very important* part is that each node determines 
> its own status, which is known universally (not like MPR, where any 
> given MPR has no universal context and hence I believe will fall 
> seriously short of scalability). Inherent robust mechanisms can also 
> elongate the lifetime of an MDR and all that an MDR-other node needs 
> to do is find its best MDR and continue to remain adjacent to it as 
> long as possible. This makes core adjacency graph as stable as 
> possible. I find this design brilliant and in the true spirit of OSPF.
>
> I have to say that this is very natural in the context of OSPF. I do 
> agree that the word *natural* can be twisted any way when taken 
> without context, but I believe when Dr. Ogier claims *natural*, he 
> thinks of scalability and a broadcast interface. Perhaps it is better 
> to call MDRs a natural way to extend broadcast interface of OSPF.
>
> Sincerely,
>
> Aniket 
>
> _______________________________________________
> Ospf-manet mailing list
> [e-mail]
> https://www1.ietf.org/mailman/listinfo/ospf-manet
>
>



_______________________________________________
Ospf-manet mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/ospf-manet