Re: [mpls] RtgDir review: draft-ietf-mpls-targeted-mldp-03.txt

[Eric Rosen <[e-mail]>]

Thank you for taking a close look at this draft again.

> 1. Section 1 is introduction, but section 1.2, 1.3, 1.4 seem to define the
> specification, not overview introduction. It is suggest to have a separate
> section to describe the introduction.

Fair enough.  Sections 1.2, 1.3, and 1.4 will be "promoted" to sections 2,
3, and 4 respectively, leaving the old section 1.1 as the introduction.

> 2. Section 1.1, the reference of mLDP RFC is indexed by [mLDP], not
> [RFC6388] (same for [LDP-UP]). I am not quite sure IETF template allows
> this, please check

This style of cross-reference is very commonly used.

> 3. Section 1.2.1, line 176, what if there is IGP neighbors, and there is
> targeted LDP session between U and D? This is possible in some
> deployments.

The cited text:

     - U is the "BGP next hop" on D's path to R, where U and D are not
       IGP neighbors, and where there is a Targeted LDP session between
       U and D.  In this case, we allow D to select U as the "upstream
       LSR" for <R,X>.

suggests that the targeted session cannot be used in this case.  Probably it
is best to leave the handling of this case as a choice for the SP, so I will
change it to ""where U and D are not necessarily IGP neighbors".  This would
allow the targeted session to be used if so chosen.

> 4. Section 1.2.1, line 188, is there a precedence among the different
> methods, and the operator could define the precedence?

I think it is best to just say (as it does) "the method is chosen by
provisioning".  Whether it is possible for an operator to configure an
ordered list of methods to try is a local matter (i.e., doesn't have to be
specified in the RFC).

> 5. Section 3, line 356, how does D know to send label request, not label
> mapping message? Multicast or unicast from U to D is determined by U, and
> D could not get this information automatically. Then is there any static
> configuration required here?

As stated in section 1.3 (section 3 in the next revision):

   "the choice between unicast replication and multicast tunneling is
   determined by provisioning, or by other methods that are outside the
   scope of this specification.  It is presupposed that all nodes will have
   a priori knowledge of whether to use unicast replication or to use
   multicast tunneling.  If the latter, it is presupposed that all nodes
   will have a priori knowledge of the type of multicast tunneling to use."

> 6. Section 3, line 358, the label request messages mentioned in the text
> should be upstream label request message. It is better to explicit to say
> this.
   
I'll add that D1 and D2 follow the procedures of Section 4 of [LDP-UP] when
requesting U to assign the label.

> 7. Section 3, line 382, it is said: While it is possible to do this, it is
> highly RECOMMENDED that MP2MP LSPs be tunneled through MP2MP LSPs. I do
> not quite understand this statement. This specification does not describe
> the mechanism of MP2MP LSPs being tunneled through MP2MP LSPs. From my
> view, MP2MP LSPs over MP2MP LSPs is not easy, and should resolve the
> upstream label uniqueness among different upstream nodes. It should be
> caution to have this recommendations here.

Perhaps it would be best just to say:

     Procedures for tunneling MP2MP LSPs through P2MP or MP2MP LSPs are
     outside the scope of this document.