The MPLS WG Archive

[Bora Akyol <akyol@pluris.com>]

I had discussed a similar point to Curtis's over the mailing list (see
archives) with Kireeti a while back. For routers that are capable of link
aggregation (IP Bond TM, Pluris marketing) some of the complex solutions are
really not required. I know that Avici does something similar as well.

If one subscribes to the UNI model, then the concept of "bonding" works very
well with the UNI model for signaling and bandwidth can be added and
subtracted successfully.

Bora


Curtis Villamizar wrote:

> In message <39CBA478.E9E457A9@tellium.com>, Bala Rajagopalan writes:
> > Curtis:
> >
> > Curtis Villamizar wrote:
> >
> > > In message <39CB7F67.BEC764D4@tellium.com>, Bala Rajagopalan writes:
> > > >
> > > >
> > > > I think we should consider the entire solution in both cases,
> > > > rather than just the bundling structure. From Yakov's mail, it
> > > > seems like the solution you propose requires:
> > > >
> > > > 1. draft-kompella-mpls-bundling-02.txt  for bundling description
> > > > 2. draft-zinin-flood-opt-00.txt for ospf flooding optimization when
> > > > there are multiple bundles between the same pair of nodes.
> > > > 3. draft-kompella-mpls-unnum-01.txt for specifying component
> > > > links in ERO
> >
> > >
> > > IMHO 2 is not required therefore your more complex bundling scheme
> > > adds litte or no value.
> >
> > This is what Yakov said:
> >
> > > (yakov) Advertising the same LSA over each control channel would be
> > > fairly unwise thing to do (to say the least). Please look
> > > at draft-zinin-flood-opt-00.txt on how to avoid doing this.
>
> Bala,
>
> I keep forgetting that our system allows multiple OC48 or OC192
> interfaces to be concatonated together and treated as a single
> interface (Composite Link TM in Avici marketing-speak) and others
> don't do this.  Therefore we do not have this a problem wrt multiple
> IGP adjacencies over the components of a bundle.  We just do SONET and
> PPP over the components and provide the appearance of a single
> interface to higher layers.
>
> I think others are doing something similar (Pluris, maybe others).
>
> Curtis