Re: [OSPF] OSPF Hybrid Broadcast and P2MP Interface Type

["Jeffrey (Zhaohui) Zhang" <[e-mail]>]

WG chairs and members,

As Acee mentioned below, this optimization is a reasonable, simple and general solution for a valid probem, and is really not conflicting with MANET.

As a result, we would like to request again for WG's acceptance of this work.

Thanks.
Jeffrey, Nischal, Lili. 

> -----Original Message-----
> From: Acee Lindem [mailto:[e-mail]] 
> Sent: Tuesday, February 22, 2011 7:10 PM
> To: Nischal Sheth
> Cc: Alvaro Retana (aretana); Jeffrey (Zhaohui) Zhang; [e-mail]
> Subject: Re: [OSPF] OSPF Hybrid Broadcast and P2MP Interface Type
> 
> Speaking as a WG member, I really don't see this hybrid 
> interface optimization as conflicting with the problem space 
> covered by the MANET draft. 
> Thanks,
> Acee 
> On Feb 10, 2011, at 6:29 PM, Nischal Sheth wrote:
> 
> > On 1/6/2011 2:26 PM, Alvaro Retana (aretana) wrote:
> > 
> >> You don't need to implement everything in the rfc to support the
> >> interface functionality.  Most of the work in the rfc is 
> oriented at
> >> reducing the overhead on the wire (Incremental Hellos, 
> Smart Peering) or
> >> at addressing the cases where not all the nodes are 
> visible (Overlapping
> >> Relays).
> >> 
> >> If you don't care about reducing the overhead and can 
> guarantee that all
> >> the nodes are visible, then the interface definition is enough. ;-)
> >> That reduces to taking advantage of the broadcast 
> characteristics for
> >> flooding, but using p2p adjacencies -- which would be a 
> lot easier to
> >> operate because it is clearer what the relationship 
> between the peers
> >> w/the different metrics is.
> >> 
> >> In my mind the problem in your document is already solved.
> >> 
> > 
> > Hi Alvaro,
> > 
> > If one were to use just the interface definition, we would 
> end up with a 
> > full mesh of adjacencies between all routers on the 
> broadcast network.
> > This is less desirable compared to the hybrid interface 
> which requires 
> > adjacencies only to the DR/BDR.
> > 
> > One would need to implement Smart Peering in order to 
> reduce the number 
> > of adjacencies on the MANET interface.  However, doing so 
> would result 
> > in suboptimal routing unless you implement Unsynchronized 
> Adjacencies.
> > Finally, Unsynchronized Adjacencies requires a protocol 
> extension which 
> > is defined only for OSPFv3.
> > 
> > Based on the points above, I don't consider the MANET 
> Interface to be a 
> > true superset of the hybrid interface to solve the problem at hand.
> > 
> > -Nischal
> > 
> 
>