Alvaro,

Sorry for the late response.

Please see below. 

> -----Original Message-----
> From: Alvaro Retana (aretana) [mailto:[email protected]] 
> Sent: Monday, December 13, 2010 3:47 PM
> To: Jeffrey (Zhaohui) Zhang
> Cc: [email protected]
> Subject: RE: [OSPF] OSPF Hybrid Broadcast and P2MP Interface Type
> 
> > From: [email protected] [mailto:[email protected]] On Behalf
> Of
> > Jeffrey (Zhaohui) Zhang
> 
> Jeffrey:
> 
> Hi!
> 
> ...
> > It is difficult to go into too much details, but there are real
> > examples of true-broadcast radio networks where the bw/delay is
> > different between different pairs of radios.
> 
> It is very tempting to use the model that you describe, but radio
> networks can be "flimsy" because of interference, signal fading,
> movement, etc..  The broadcast model for neighbor 
> discovery/maintenance
> can quickly become unusable for nodes that loose connectivity 
> to the DR.
> It is very easy to see how the same L2 media may provide nodes that
> don't have the same view of the other nodes (not everyone is within
> radio range of everyone else, for example).

There are real-life examples where a radio network is built to guarantee that 
if a node can reach a central node then it can reach others who can also reach 
the central node (e.g., a satellite network). A node's transimission/receiving 
capability can be impacted by its placement, but the broadcast property of the 
network is still maintained (if a station is so out of place then it is 
considered off the network, just like a cable is cut in the wired ethernet 
case).

The premise is that we have a broadcast network but one can reach some stations 
with a metric that is different from when reaching others.

If that premise is not satisfied, then it's a different topic (and out of the 
scope).

Thanks.
Jeffrey

> 
> This brings me to wanting to establish p2p adjacencies while using the
> broadcast capabilities of the media to reach whoever is within
> transmission range (to propagate Hellos and to flood LSAs).  The
> recognition that the characteristics between nodes can be different
> should also result in the recognition that the adjacencies 
> between pairs
> of nodes may not all have the same state at the same time.
> 
> My 1c.
> 
> Alvaro.
> 
> 
_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf

Reply via email to