RE: I-D ACTION:draft-andersson-mpls-g-chng-proc-00.txt

[Shahram Davari <Shahram_Davari@pmc-sierra.com>]

>-----Original Message-----
>From: Curtis Villamizar [mailto:curtis@fictitious.org]
>Sent: Wednesday, March 05, 2003 3:27 PM
>To: Shahram Davari
>Cc: 'Mark.Jones@mail.sprint.com'; ccamp@ops.ietf.org; mpls@UU.NET
>Subject: Re: I-D ACTION:draft-andersson-mpls-g-chng-proc-00.txt 
>
>
>
>In message 
><4B6D09F3B826D411A67300D0B706EFDEB03D60@nt-exch-yow.pmc-sierra.bc.ca
>>, Shahram Davari writes:
>> >Regardless of the process that the IETF uses to address the 
>> >contributions, network need for extensions, to address applications 
>> >outside the IETF scope, will continue to be identified.  For that 
>> >reason, there will be real network need for those extensions.  
>> 
>> It might be a good idea to require that all IETF protocols support
>> vendor-specific extensions, so that they could be used by other SDOs
>> and for experiments.
>> 
>> -Shahram
>
>
>How does a committee experiment with a protocol?  AFAIK most SDOs
>don't "experiment" they just write standards without trying them out,
>then try to mandate that everyone implement them still not knowing for
>sure if they actually work.  Its the IETF that experiments first.
>That's the "running code" thing you might have heard about.

I didn't say the SDOs want to experiment. I said the vendor-specific
could be used by both SDOs and those who want to experiment.

>
>All of the protocols support extension mechanisms.  One or more
>vendors often experiment with protocol extensions, label documents
>with TBD for codepoints and pick an unused one temporarily.  Once
>sufficient consensus is achieved that the protocol extension is at
>least worthwhile, which often follows successful limited deployment,
>the TBD is given an IANA assignment while the details are worked out.
>This works just fine.

Who guarantees that the very codepoint they are using will not 
be standardized by IETF for other use, while they are experimenting it?
And why LDP has such extension? 
  
>
>Running code and successful deployment do carry weight in the IETF but
>even that is not absolute as evident by the MPLS/ICMP extensions which
>were successfully deployed but rejected.
>
>If you mean that any committee should be able to try to mandate any
>extension they please, then I don't think its a good idea.


I think IETF should provide vendor-specific extension (like the one in LDP),
and that other SDOs need to be free to use it for any purpose they want. If
their idea is not good, people won't use their extension, period. No harm is done 
to the protocol.

-Shahram