The MPLS WG Archive

[Dirk Ooms <Dirk.Ooms@alcatel.be>]

Ali Boudani wrote:
> 
> Hello,
> 
> DIRK
> Do you mean : that the best route (who's maybe not the shortest one due to
> some constraints) is considered as the RPF now

I think you got it, but I want to avoid the word 'best'.  In the
multicast topology, there will be a route to the source and this route
is taken into account for the RPF (this can be the shortest route
calculated by e.g. OSPF, but it can also be a route calculated by a more
sophisticated routing algorithm).

dirk

dirk

> 
> and vijay
> what exactly do you mean about your question ??
> 
> Thanx
> Ali
> 
> Vijay Gill wrote:
> 
> > On Wed, 10 Jan 2001, Ron Bonica wrote:
> >
> > > 2) If RPF checks succeed only when traffic arrives on the interface
> > > that IGP metrics dictate, what is the benefit of running IP multicast
> > > over MPLS?
> >
> > Raising a heretical thought here, has someone raised the (operational)
> > solution of just providing an overlay multicast network by emulating a
> > layer 2 network over mpls and running multicast over that? This also
> > manages to eliminate a possible S,G state explosion affecting the current
> > revenue generating unicast network. [1]
> >
> > Not a very clever solution, but has the added benefit of working without
> > adding more protocol fragility into already broken system implementations.
> >
> > /vijay
> >
> > [1] credit must be given to Giles Heron (L3) for this.