The MPLS WG Archive

[Bora Akyol <akyol@pluris.com>]

I would agree with Eric here.

Since I started this thread, about three alternate (maybe complementary
approaches) to doing hierarchical tunnels with RSVP-TE got proposed:

1) Tony Li's email suggested that one can use RSVP-TE by tunneling the
RSVP messages between A and F via the tunnel between B&E.

2) There was a suggestion of using directed LDP over the first tunnel to
establish the second tunnel.

3) And finally a reference to draft-kompella-*** came out, which I admit
I haven't read yet, but will probably do tonight.

I think this topic deserves farther discussion. We are NOT writing a
research paper here, but trying to produce working, interoperable code.
So those that have this working in a network, lab, demo etc; please come
forward and speak up.

Do we need to have an informational ID on this?

Bora


Eric Gray wrote:

> Shahram/Tony,
>
>         I think we may be glossing over some of the
> complexities in this discussion.
>
>         In the diagram given earlier:
>
> A -- B -.           .- E -- F
>          `- C -- D -'
>
> We want to create an outer tunnel from B to E to
> transport MPLS labeled packets in the inner tunnel
> from A to F.  Two sets of RSVP-TE messages would be
> required: those going from A to F (the inner tunnel)
> and those going from B to E (the outer tunnel).  The
> RSVP-TE messages associated with the inner tunnel
> would be transported using the labels associated with
> the outer tunnel in going from B to E. The RSVP-TE
> messages associated with the outer tunnel would be
> either transported using raw IP or using LSP(s) with
> E (and B) as LSP destinations - in other words, yet
> another set of LSPs.
>
>         For RSVP-TE messages being used to support the
> inner tunnel, Shahram is correct - no special label
> is needed by E.  This is because E is the egress for
> the outer tunnel and either D will pop the label for
> the outer tunnel, or E will.  Whether D is asked to
> do a PHP for the outer tunnel or not depends on the
> capabilities of D as an LSR/LER.  Whether or not an
> explicit NULL label is required by E depends on the
> capabilities of E.  For example, for some traffic
> going through the outer tunnel, E may choose to ask
> B to do a PHP for an inner tunnel (yet another one)
> that - for instance - goes from A to E.  This would
> also produce IP encapsulated packets inside the outer
> tunnel LSP.  If LSR D is doing PHP for the outer
> tunnel, it must be able to distinguish when it is
> popping the last label and when it is not so that it
> will produce the correct L2 encapsulation for the
> link from D to E.  Once it has done this, then E will
> be able to demux labeled and unlabeled traffic
> received from D.
>
>         Both simpler and more complex than discussion
> seemed to make it out to be...
>
> --
> Eric Gray
>
> > -----Original Message-----
> > From: Shahram Davari [mailto:Shahram_Davari@pmc-sierra.com]
> > Sent: Monday, June 12, 2000 3:13 PM
> > To: 'Tony Li'; Bora Akyol
> > Cc: 'mpls@uu.net'
> > Subject: RE: Hierarchical Tunnel Establishment in RSVP-TE
> >
> >
> > Hi Tony,
> >
> > Since an S-bit=0 indicates that the encapsulated packet is an
> > MPLS packet,
> > can't we conclude that L3PID is used only for the
> > identification of L3? In
> > other words it is possible to tunnel both labeled and
> > unlabeled packets
> > through the same MPLS tunnel (i.e., with the same outer label)?
> >
> > If this is true then you don't need to set the L3PID=MPLS for
> > the outer
> > tunnel. Instead you can set the L3PID=IP and use the same
> > tunnel for the
> > tunneled MPLS packets as well as tunneled RSVP messages.
> >
> >
> > Regards,
> > -Shahram
> >
> > >-----Original Message-----
> > >From: Tony Li [mailto:tli@Procket.com]
> > >Sent: Saturday, June 10, 2000 1:25 AM
> > >To: Bora Akyol
> > >Cc: 'mpls@uu.net'
> > >Subject: Hierarchical Tunnel Establishment in RSVP-TE
> > >
> > >
> > >
> > >
> > > | How does one establish hierarchical tunnels using RSVP-TE?
> > > |
> > > | I know that the label object can be deeper than one label,
> > >but you need more
> > > | than this to establish hierarchical tunnels.
> > >
> > >
> > >A hierarchical tunnel requires the participation of at least
> > >two LSRs.  The
> > >operation of the LSR creating the innermost tunnel is
> > >straightforward and
> > >is unchanged from what has been endlessly discussed here.
> > >
> > >For an LSR to create an outer tunnel, it simply creates
> > >another tunnel and
> > >specifies MPLS as the L3PID.  It then forwards packets
> > received on the
> > >inner tunnel by pushing a label for the outer tunnel.  The
> > RSVP control
> > >traffic for the inner tunnel is also forwarded through the
> > outer tunnel
> > >after first being encapsulated with a null IP label.
> > >
> > >Recurse, ad nauseum.
> > >
> > >Simple, really.  ;-)
> > >
> > >Tony
> > >
> >