RE: [Ospf-manet] URL for MPR-extension software

Emmanuel, some responses inline below:

> -----Original Message-----
> From: Emmanuel Baccelli [mailto:Emmanuel.Baccelli@inria.fr] 
> Sent: Monday, December 18, 2006 12:51 AM
> To: ospf-manet@ietf.org
> Subject: RE: [Ospf-manet] URL for MPR-extension software
> 
> 
> Tom,
> 
> We are saying that topology reduction is very effective. 

The above point is not in dispute.  But previously, you were saying:

> We showed during our last presentation at the IETF that topology 
> reduction is at least as effective as adjacency reduction, if 
> not more. 
> This is the point that was made, based on the results we published.

This point doesn't seem to be substantiated, and is what I was asking
for clarification.  

> In fact, the
> load of the full LSAs is much larger than the load due to
> synchronization, as seen with all the simulations published so far.
> 

While the above might be technically true, the cost of adjacencies is
not just synchronization overhead but additional Ack traffic (as Richard
pointed out).

> With the MPR topology reduction, up to 50% of the load due to LSAs can
> be avoided. It is difficult to match that with adjacency reduction on
> its own. However, adjacency reduction can indeed be 
> complimentary, and 
> this is also possible with MPR-OSPF, as described in the 
> draft (instead 
> of bringing up adjacencies with all its wireless neighbors a wireless
> router MAY bring up adjacencies with some of them), while still being
> OSPF compliant in the sense that routing is done over 
> adjacencies only.
> 

The main reason that I became interested in MDR after our initial start
with an MPR-based solution was the realization that MPRs were likely to
be less useful as a means of adjacency thinning, because there are
likely to be many more MPR/selector relationships than MDR-formed
adjacencies.  I am not really satisfied by rough analytical estimates
that adjacency reduction would be much less valuable than topology
reductions because these estimates seem to undercount the actual
transmissions that arise when you run the full-blown protocol.
Therefore, it seems to me that the potentially interesting detail to
look at more carefully (the effectiveness of MPR-based adjacency
reduction) is unfortunately missing from your code.  

Tom

_______________________________________________
Ospf-manet mailing list
Ospf-manet@ietf.org
https://www1.ietf.org/mailman/listinfo/ospf-manet