Question on tunnel Interface (w.r.t TE MIB)

-----Original Message-----
From: Thomas Nadeau [mailto:tnadeau@cisco.com]
Sent: Monday, October 29, 2001 12:05 PM
To: Dominic-Savio, Patrick
Cc: 'mpls@uu.net'
Subject: Re: Question on tunnel Interface (w.r.t TE MIB)

Lets first be specific about which document
we are referring to. It sounds like you are referring to
the MPLS-TE-MIB. If you are, then read on, otherwise
skip my reply.

Sorry I was not specific, but yes, I am referring to the MPLS-TE-MIB.

Here is my question:
When you create a tunnel at the head-end, you can set the mplsTunnelIsIf
variable to TRUE and this specifies that an interface will be created over
the tunnel at the head-end.

Not really. This means that the tunnel IS the ifIndex associated
with it.

What is the procedure for identifying at the
tail-end that this tunnel is to be used as an interface (for the purpose of
receiving packets coming from the head-end)?.

First, note that this functionality is optional in the MIB.

Even if it is optional, I think it should be clearly specified how the feature is to be supported using the MIB table(s).

Second,
this can be implemented only if your device creates an tunnel entry at
the egress.

Yes, of course.

Note that this will be at the penultimate-hop, if that is
enabled.
Why should it be at the penultimate-hop ?. RSVP signalling goes all the way to the tail-end and the tail-end TE tunnel entry can be created there. (Also, penultimate-hop popping is optional).

Should a tail-end manual entry be made in the TE table ahead of time with
the mplsTunnelIsIf value set to TRUE?.

        Depends on your implementation. I suppose that you could manually
create such an entry, but I am not sure why. Some implementations represent
the tail of the tunnel the same way that other midpoints are represented (without
an ifIndex) except that the tunnelRole would be set to tail(3).

The reason to create this entry would be so that the user can set the  mplsTunnelIsIf value to TRUE. So it is clear based on the configuration that this terminating tunnel will be a terminating point of an unidirectional interface.

If so, it does not seem like there is
a (documented) way to uniquely identify an incoming signalling message (for
tunnel termination) using the tail-end TE MIB specification. In other words
how do I match a user created tail-end configuration with (say) a RSVP
session that is terminating on this tail-end LSR.

You should be able to figure this out given the RSVP-TE information
stored at the tail-end router. It should tell you that this RSVP-TE session
terminates at that LSR the same way that a mid-point LSR tells keeps
information about transit-sessions.

So you are saying that the terminating point of a tunnel interface is identified on the terminating router based on the RSVP session information for that tunnel?. But that is confusing if the user has to use the mplsTunnelTable to configure interface capability on the head-end and use some other table to configure the interface capability at the tail-end.

Let me give you an example, lets say you have router A and router B, with some intermediate hops (that are of no consequence).

There is a tunnel (tunnel index 1) with mplsTunnelIsIf set to TRUE originating on Router A and terminating on Router B. Also mplsTunnelName for tunnel 1 is "tunIntf1"

So you have an interface by the name of "tunIntf1" on Router A.

If you want to have an interface on Router B to receive packets sent from Router A on tunIntf1, how do you set it up ? and where do you specify the tunnel interface name on Router B?

Thanks

--Tom

------------------------------------------------------------------------

Mathematics is the supreme nostalgia of our time.

The MPLS WG Archive

Question on tunnel Interface (w.r.t TE MIB)