[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Diffserv] Re: Problems with IP-Multicast in DiffServ-Networks

Steve Deering wrote:

> What I saw was Brian's comment about "a 10-party conference needs to be
> provisioned on a multipoint basis", and Albert drawing analogies to
> ATM pt-to-mpt flows, which made me think you were talking about
> per-(micro)flow reservations.  That's what I get for dropping in
> without sufficient context.
> By then why are you discussing "reservations" at all?  Isn't that the
> domain of int-serv?

I think i have a terminology problem here.

> > > any router along the delivery tree may modify
> > > those bits to indicate how the packet ought be handled from that point
> > > onward, along one or more descendent branches of the delivery tree.
> >
> >i'm not clear  how this will work in a diffserv network. Could you elaborate
> >pls?
> I would elaborate if I understood what you think the problem is.  You
> provision resources in your networks for certain amounts of traffic
> labelled in certain ways.  Multicast packets carrying such labels consume
> the corresponding resources along the branches of their delivery trees,
> and are accounted for, just like unicast packets that happen to flow along
> the same paths.

Yes i understand your explanation above.
But i 'm not sure of modifying bits in interior /core routers.

> >How to provision for multicast groups?
> You don't provision for multicast groups.  You provision for aggregate
> flows of traffic, to a potential multiplicity of destinations, both
> unicast and multicast.

Yes, i see the misunderstanding, i meant provisioning for multicast groups for
aggregated traffic which was what i think we are discussing. Actually the
proposal i posted allows provisioning for aggregate flows of traffic, but
leverages multicast tree construction to do this.


diffserv mailing list
Archive: http://www-nrg.ee.lbl.gov/diff-serv-arch/