The MPLS WG Archive

["Phelan, Tom" <tphelan@sonusnet.com>]

Hi Eric,

See inline ...

Tom Phelan

> -----Original Message-----
> From: Eric Rosen [mailto:erosen@cisco.com]
> Sent: Thursday, October 21, 2004 3:25 PM
> To: mpls@ietf.org
> Subject: [mpls] Host Address FEC
> 
> 
> We  had originally  agreed to  remove  the Host  Address FEC  
> on grounds  of
> disuse, but this  decision was deferred due to  a liaison 
> statement received
> from some forum which claims  to have written specifications 
> which depend on
> it. 
> 
> However, I find  that liaison statement very confusing.   In 
> particular, the
> liaison states: 
> 
>     "In RFC 3036, there is a  semantic difference between a 
> Host Address and
>     a  prefix  with length  32.   The new  version  proposes  
> to remove  the
>     semantic difference.  This is not a problem for the MPLS 
> Proxy Admission
>     Control; we are  just requesting that the Host  Address 
> codepoint not be
>     deprecated."
> 
> This seems to be  saying that they want the Host Address  FEC 
> to remain, but
> they don't  care if its semantics  are changed.  That  is 
> somewhat peculiar.
> To me, it suggests that they are  not actually using the Host 
> Address FEC as
> it is specified  in RFC 3036, but  rather are overloading it 
> to  be used for
> another purpose entirely;  in effect stealing the codepoint  
> for a different
> use. One might think that the  very fact that this codepoint 
> is stealable is
> itself an indication of its lack of use.

No, you've got it backwards.  We want the Host Address FEC to remain, with its current syntax.  The MPLS Forum protocol uses the Host Address FEC as a host address, as it's intended in RFC 3036.  The proposed semantic change was to the Prefix Address FEC (not the Host Address FEC) - to make a Prefix Address with a 32-bit prefix equivalent to a Host Address FEC.  This change, to make a Prefix Address FEC able to substitute for a Host Address FEC has no effect on our protocol.  The MPLS Forum protocol needs to have the Host Address FEC as it is in RFC 3036, and uses it in the spirit of its meaning there.

> 
> It seems to me then that we should just go ahead and remove 
> the Host Address
> FEC from rfc2026bis.  If some other organization needs a TLV 
> whose usage has
> not  been specified  in  any IETF  document,  they should  
> follow the  usual
> procedures and write  an internet draft requesting a  code 
> point assignment.
> If they want  to reuse an existing codepoint in  the "IETF 
> consensus" range,
> they should obtain IETF consensus. 

Well, what we actually need is a TLV whose usage *has* been specified in an IETF document, and would like to see it continue to be specified.  Something vastly different than what you're saying here.

_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls