The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2004-Mar> msg00042



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

query on draft-rosen-mpls-explicit-null-01

  • From: Eric Rosen <erosen@cisco.com>
  • Date: Fri, 12 Mar 2004 11:05:59 -0500
  • cc: "Loa Andersson" <loa@pi.se>, "MPLS wg" <mpls@UU.NET>, "Alex Zinin" <zinin@psg.com>, "George Swallow" <swallow@cisco.com>
  • User-Agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.3(Unebigoryōmae) APEL/10.3 Emacs/21.3(sparc-sun-solaris2.8) MULE/5.0 (SAKAKI)

Adrian> In the 3032  definition, it is possible to tell  from the label what
Adrian> further  action  must  be taken  (viz,  look  for  an IPv4  or  IPv6
Adrian> address).  In your  proposal it  is necessary  to examine  the label
Adrian> _and_ the  label stack bit to  determine whether to act  on the next
Adrian> label or on the IP address.  There is nothing wrong with this, 

Good.  

Adrian> but I  wondered if you had  considered a third  value "MPLS Explicit
Adrian> NULL label" to  cover your requirements so that  only one item needs
Adrian> to be examined?

That  wouldn't  work.   When you  receive  an  MPLS  packet, the  top  label
determines the  label stack operation, e.g.,  replace label X  with label Y.
There is no  provision for saying "replace label  X with label Y if  X is at
the bottom of the stack, but with label Z otherwise".

Adrian> On the  other hand,  why don't we  support explicit null  labels for
Adrian> other payloads?  

As this question is not about a change from RFC 3032, it is out of scope for
this discussion ;-) 

Adrian> Document  header  should probably  include  the  key words  "updates
Adrian> RFC3032"

Good idea, done.

Adrian> Abstract refers to RFC 3032 without giving full details. 

I  think  the RFC  editor  rules  are to  avoid  citations  in the  abstract
altogether, so I will rewrite the abstract.

Adrian> In my mind "subsequent" is  a difficult word since some might regard
Adrian> the  label to  use as  "previous".  Can we  say "next  label in  the
Adrian> stack"? 

But "next" and  "subsequent" mean the same thing.  If  you really think this
is confusing, I  will say "the label beneath it in  the stack", or something
like that.

Adrian> You also refer to RFCs 2026, 3031 and 3036. 

I will add them to the references.

Adrian> Can we have the (new) IPR statement and copyright notice before this
Adrian> becomes a WG draft, please. 

Added to the document source;  it will be there in draft-ietf-mpls-explicit-
null-00, if the WG approves this as a WG doc.