The MPLS WG Archive

[abhay <abhay@multitech.co.in>]

Tom,

    This new explanation would have helped me sometime ago,

    It did take a while for me to get  what you were saying then.

    Better late than never..8)

    Thanks

    Regards,

    Abhay

"Thomas D. Nadeau" wrote:

>In section 7 of draft-ietf-mpls-te-mib-06.txt the interface
>stack is shown as:
>
>         +------------------------------------------------+
>         | MPLS-Tunnel-interface ifType = mplsTunnel(150) |
>         +------------------------------------------------+
>         |        MPLS-interface ifType = mpls(166)       |
>         +------------------------------------------------+
>         |               Underlying Layer                 |
>         +------------------------------------------------+
>
>It is not clear to me how the particular lower layer of the tunnel
>interface is chosen.
>
>In the case of a tunnel interface created but not yet signalled, I
>imagine it would be layered over nothing (ifindex 0).
>
>Once signalled, would it be the mplsOutSegmentIfIndex of the
>tunnel's underlying cross connect?
>
>If so, I assume a rerouted tunnel, since it might have a different egress
>interface, might need to be restacked onto a different interface.
>
>Should a tunnel also be layered over the mplsInSegmentIfIndex (if
>nonzero)?

         Sorry for the confusion. The text in the draft needs
a few sentences to clarify. Something like this:

         In the case where the tunnel interface is
present at the head-end of the tunnel, the
tunnel interface should be stacked upon the
outgoing MPLS interface.  In the case where
the tunnel interface is optionally represented
at the egress of the tunnel as well, it
should be stacked upon the incoming MPLS
interface. If the tunnel interface has been created,
but no LSP has been signaled for the tunnel,
the tunnel should not be stacked upon any underlying
interface. If the tunnel's underlying LSP is
changed, then the interface stacking should be
changed accordingly to the new MPLS interface
as described above.

         --Tom