The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2006-Apr> msg00033



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

[mpls] New Liaison Statement, "T-MPLS Consented Recommendatio ns"

  • From: "Thomas D. Nadeau" <tnadeau@cisco.com>
  • Date: Tue, 11 Apr 2006 12:16:31 -0400
  • Cc: mpls@lists.ietf.org
  • X-Brightmail-Tracker: AAAAAA==
  • X-IronPort-AV: i="4.04,112,1144047600"; d="scan'208"; a="25656870:sNHT30160728"
  • X-OriginalArrivalTime: 11 Apr 2006 16:16:30.0830 (UTC)FILETIME=[4B1368E0:01C65D83]


On Apr 11, 2006, at 11:48 AM, <neil.2.harrison@bt.com> wrote:

>> 	How do you trace the LSP using Y.1711 or is that
>> not an important function for OAM for T-MPLS? Also, from the
>> IETF's perspective, we have already defined a tool set that
>> satisfies all of the requirements of ping and trace (tree and
>> path) in a single tool, which is now on the standards track,
>> is widely deployed, and thus should be the tool specified herein.
>
> Tom, tracing in a co-ps mode technology that respects connection
> requirements should be a whole lot easier than in one than does  
> not.  So
> I would argue that you can re-use the same diagnostic tools in T-MPLS.

	Theory and practice are very different when it comes
to data plane diagnostics vs. control plane. Hoping that it
is easier and ensuring that it actually works %100 of the time
are two quite different things.  Also, as someone who works at a
network operator, you surely can appreciate the advantages of using
a single mechanism for fault-detection and trouble-shooting,
especially if it is widely deployed and available.

	--Tom


> BTW - We really should start with a definition of defects including
> entry/exit criteria...everything else builds on this baseline.  Unless
> one has a proper connection however, I really don't see how this is
> possible.
>
> regards, Neil
>>
>> 	--Tom
>>
>>
>>> -Shahram
>>>
>>>> -----Original Message-----
>>>> From: Thomas D. Nadeau [mailto:tnadeau@cisco.com]
>>>> Sent: Tuesday, April 11, 2006 10:34 AM
>>>> To: Shahram Davari
>>>> Cc: mpls@lists.ietf.org
>>>> Subject: Re: [mpls] New Liaison Statement, "T-MPLS Consented
>>>> Recommendatio ns"
>>>>
>>>>
>>>>
>>>>> Hi Tom,
>>>>>
>>>>> 1) Reading the T-MPLS specs I didn't see any specific label range
>>>>> restriction. Where have you seen this?
>>>>
>>>> 	See inline marked with "TOM:"
>>>>
>>>>> 2) The OAM specified is for T-MPLS, which as you said is
>> a subset of
>>>>> MPLS. Specifically T-MPLS doesn't include PHP, merging
>> and ECMP as
>>>>> indicated by the Liaison. So why should an OAM that is
>> going to be
>>>>> used for T-MPLS support those functions?
>>>>
>>>> 	The OAM as specified in the T-MPLS document is not a
>> subset of MPLS;
>>>> it is a completely different mechanism from what is
>> specified at the
>>>> IETF to meet the requirements in RFC4377. As described,
>> y1711 acts on
>>>> a sub-set of MPLS (lets forget about the label range), but
>> does not
>>>> meet all of the requirements as specified in RFC4377, even
>> for this
>>>> subset of MPLS.
>>>>
>>>> 	--Tom
>>>>
>>>>
>>>>
>>>>>
>>>>>
>>>>> Regards,
>>>>> -Shahram
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Thomas D. Nadeau [mailto:tnadeau@cisco.com]
>>>>>> Sent: Tuesday, April 11, 2006 9:25 AM
>>>>>> To: mpls@lists.ietf.org
>>>>>> Subject: Re: [mpls] New Liaison Statement, "T-MPLS Consented
>>>>>> Recommendations"
>>>>>>
>>>>>>
>>>>>> [Resending to mpls list as I got a warning from the mailer that
>>>>>> there was a suspicious number of recipients, so I am not sure it
>>>>>> was delivered.]
>>>>>>
>>>>>> 	Hi,
>>>>>>
>>>>>> 	A few things in the ITU T-MPLS document concern me:
>>>>>>
>>>>>> 	1) The document itself constitutes what is essentially
>>>>>> 		two things: an protocol specification
>>>>>> 		and an implementation guideline.
>>>>>>
>>>>>> 		For this reason, I first suggest that this document
>>>>>> 		needs to be split into two: one that specifies
>>>>>> 		the label range allocation (as well as requests
>>>>>> 		it from the IETF), and a second that specifies
>>>>>> 		the implementation guideline. I also suggest
>>>>>> 		removal of "MPLS" from the name. This is very
>>>>>> 		confusing, and implies that this document is
>>>>>> 		some new standard for MPLS, whereas what it
>>>>>> 		really specifies (sans the label range allocation
>>>>>> 		and the OAM) is a subset of MPLS functionality.
>>>>>>
>>>>>> 	2)	With regard to the label range reservation, I am
>>>>>> 		troubled by this because MPLS has been widely deployed
>>>>>> 		without the said reserved label range. Deployment
>>>>>> 		of LSRs that suddenly use this special label
>>>>>> 		range will most likely lead to interoperability
>>>>>> 		issues.  I am also troubled by the lack of
>>>>>> 		definition of what the reserved labels should
>>>>>> 		be used for.  If in the future, these labels are
>>>>>> 		defined in a manner that is incompatible with
>>>>>> 		existing LSRs, this will pose a significant
>>>>>> 		hurdle to deployments. Again, a separate document that
>>>>>> 		requests the label range, as well as specifies
>>>>>> 		what they are to be used for is appropriate.
>>>>>>
>>>>>> 	3) 	The OAM as specified does not meet the IETF's
>>>>>> 		requirements for OAM specified in RFC4377.
>>>>>> 		Y.1711 is not suitable for environments where PHP is
>>>>>> 		enabled, and it also has other deficiencies such as a
>>>>>> 		lack of tracing functionality, and cannot handle Equal
>>>>>> 		Cost Multi-Path (ECMP) LSPs.  In particular, section
>>>>>> 		4.3 of RFC4377 stipulates that path characterization
>>>>>> 		functions must be able to work in these cases:
>>>>>>
>>>>>> 4.3.  Path Characterization
>>>>>>
>>>>>>     The path characterization function is the ability to reveal
>>>>>> details
>>>>>>     of LSR forwarding operations.  These details can then
>>>> be compared
>>>>>>     during subsequent testing relevant to OAM
>> functionality.  This
>>>>>>     includes but is not limited to:
>>>>>>
>>>>>>        -  consistent use of pipe or uniform time to live
>>>>>> (TTL) models by
>>>>>>           an LSR [RFC3443].
>>>>>>
>>>>>>        -  sufficient details that allow the test origin
>> to exercise
>>>>>> all
>>>>>>           path permutations related to load spreading
>> (e.g., ECMP).
>>>>>>
>>>>>>        -  stack operations performed by the LSR, such as pushes,
>>>>>> pops,
>>>>>>           and TTL propagation at penultimate hop LSRs.
>>>>>>
>>>>>> 	I suggest replacing the reference to Y.1711 with a reference
>>>>>> 	to rfc4377, rfc4378, and rfc4379.
>>>>>>
>>>>>> 	--Tom
>>>>>>
>>>>>>
>>>>>>
>>>>>>> Title: T-MPLS Consented Recommendations
>>>>>>> Submission Date: 2006-04-02
>>>>>>> URL of the IETF Web page: https://datatracker.ietf.org/public/
>>>>>>> liaison_detail.cgi?detail_id=207
>>>>>>> Please reply by 2006-04-17
>>>>>>>
>>>>>>> From: Greg Jones(ITU-T SG 15) <tsbsg15@itu.int>
>>>>>>> To: IETF MPLS WG, CC: MFA Forum(George Swallow
>>>>>> <swallow@cisco.com>,
>>>>>>> Loa Andersson <loa@pi.se>)
>>>>>>> Cc: info@mfaforum.org
>>>>>>> maeda@ansl.ntt.co.jp
>>>>>>> sjtrowbridge@lucent.com
>>>>>>> rcallon@juniper.net
>>>>>>> fenner@research.att.com
>>>>>>> sob@harvard.edu
>>>>>>> sbrim@cisco.com
>>>>>>> mpls@lists.ietf.org
>>>>>>> Reponse Contact: tsbsg15@itu.int
>>>>>>> Technical Contact: Ghani.Abbas@marconi.com mark.jones@sprint.com
>>>>>>> betts01@nortel.com
>>>>>>> Purpose: For action
>>>>>>> Body: We apologize for not having sent a liaison regarding three
>>>>>>> consented Recommendations related to
>>>>>>> T-MPLS. The intent of these Recommendations is not to
>> change MPLS
>>>>>>> but rather to identify a
>>>>>>> subset of existing MPLS necessary and sufficient to provide
>>>>>>> connection-oriented packet transport.
>>>>>>> The approach to the work was to make a profile of the MPLS data
>>>>>>> plane functionalities defined in
>>>>>>> IETF RFCs using the description of MPLS provided in
>> G.8110 (which
>>>>>>> was approved last year).
>>>>>>> The intended application of this profile is to provide a
>>>>>> connection-
>>>>>>> oriented packet transport
>>>>>>> network.
>>>>>>> The focus of the three new recommendations is on the data plane
>>>>>>> aspects of T-MPLS. Control plane aspects are currently
>> for further
>>>>>>> study and the work on this subject is starting.
>>>>>>> According to our work methodology the T-MPLS
>> definitions are split
>>>>>>> into the three main
>>>>>>> Recommendations that have been consented in February 2006:
>>>>>>> *	G.8110.1 dealing with the T-MPLS Architecture
>>>>>>> *	G.8112 dealing with T-MPLS interface specification
>>>>>>> *	G.8121 dealing with T-MPLS equipment functional
>> specification
>>>>>>> A brief high-level introduction of the essential features and
>>>>>>> selected options for Transport MPLS
>>>>>>> can be found in section 6/G.8110.1.
>>>>>>> The objective of the work done by ITU-T SG15 was to define
>>>>>> a way to
>>>>>>> use MPLS as a connection-
>>>>>>> oriented packet-based transport solution for packet aware
>>>>>> transport
>>>>>>> networks that can support both
>>>>>>> packet and circuit (e.g. SDH, OTH or WDM) switching
>> technologies
>>>>>>> under a common operation, control and management paradigm. For
>>>>>>> this purpose T-MPLS deploys the MPLS frame format,
>>>>>>> Client to MPLS mapping and MPLS to MPLS multiplexing and
>>>>>>> complements this with transport
>>>>>>> network OAM (Y.1711), nested connection monitoring requiring the
>>>>>>> labels to be present up to the
>>>>>>> final node in the network, protection switching (Y.1720/G.8131),
>>>>>>> transport network based control
>>>>>>> and management planes, maintaining frame ordering and a
>> restricted
>>>>>>> number of classes of
>>>>>>> service/queues.
>>>>>>> *	MPLS has many applications and it is desirable
>> to identify a
>>>>>>> subset of MPLS technology that
>>>>>>> provides the necessary and sufficient functionality for
>> transport
>>>>>>> applications.
>>>>>>> We have indicated below some of the options we have selected :-
>>>>>>> *	In IETF RFCs the usage of PHP is optional. We decided
>>>>>> not to use
>>>>>>> it in order to simplify OAM
>>>>>>> procedures
>>>>>>> *	In IETF RFCs the usage of merging is optional. We
>>>>>> decided not to
>>>>>>> use it in order to simplify
>>>>>>> OAM procedures and also because the relative lower
>> number of LSPs
>>>>>>> to be supported is not considered a major scaling issue
>>>>>>> *	The usage of ECMP is optional. We decided not to use it
>>>>>> in order
>>>>>>> to simplify OAM procedures
>>>>>>> *	Leaving labels 16-31 for further study was seen
>> as a no change
>>>>>>> because the label allocation on a
>>>>
>>>> TOM: Read above.
>>>>
>>>>>>> link is defined in IETF RFCs as a local matter for any
>>>> MPLS device.
>>>>>>> We are not expecting that the current version of the
>> specification
>>>>>>> is going to create any interoperability issue. We envisage two
>>>>>>> possible cases of
>>>>>>> interoperability:
>>>>>>> 1.	Interoperability between a T-MPLS box and an
>>>> existing full-
>>>>>>> featured and fully configurable
>>>>>>> MPLS box can be solved by configuring the MPLS box to
>> use the same
>>>>>>> options selected by T- MPLS (T-MPLS being a profile of
>> MPLS this
>>>>>>> should be possible). In this case the link between
>>>>>>> the two boxes is a T-MPLS link and in the scope of our
>>>>>>> Recommendations.
>>>>>>> 2.	Interoperability between a transport platform
>>>>>> supporting T-MPLS
>>>>>>> and an existing MPLS box
>>>>>>> that uses a different option selection than T-MPLS, should be
>>>>>>> solved by the transport platform. In this case the link between
>>>>>>> the transport platform and the MPLS box is not a T-MPLS link and
>>>>>>> therefore it is outside the scope of T-MPLS
>>>>>> recommendations. We are
>>>>>>> looking for cooperation
>>>>>>> with IETF to develop the detailed specification of this case.
>>>>>>> Please find attached the text of the consented Recommendations.
>>>>>>> Please note that some minor modifications may be made to these
>>>>>>> draft Recommendations as a result of comments made during
>>>>>>> the approval process
>>>>>>> We will appreciate your comments and suggestions in
>>>>>> relation to the
>>>>>>> initial scope of T-MPLS as a
>>>>>>> packet transport network technology.
>>>>>>> We are planning to review your comments during the next meeting
>>>>>>> that is planned in Kobe (Japan) on 22-27 April 2006. Results of
>>>>>>> the comments will be included into our future work (e.g.
>>>>>>> corrigenda or amendments) on T-MPLS Recommendations.
>> Amendments of
>>>>>>> T-MPLS Recommendations are
>>>>>>> already in our plan for the next SG15 plenary meeting
>> in November
>>>>>>> 2006.We are looking forward
>>>>>>> for future cooperation with you in T-MPLS data plane and control
>>>>>>> plane evolution.
>>>>>>>
>>>>>>> Attachments: Consented Text of G.8110.1, G.8112, G.8121
>>>>>>> Attachment(s):
>>>>>>>      Consented Text of G.8110.1 (https://datatracker.ietf.org/
>>>>>>> documents/LIAISON/file288.zip)
>>>>>>>      Consented Text of G.8112 (https://datatracker.ietf.org/
>>>>>>> documents/LIAISON/file289.zip)
>>>>>>>      Consented Text of G.8121 (https://datatracker.ietf.org/
>>>>>>> documents/LIAISON/file290.zip)
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> mpls mailing list
>>>>>>> mpls@lists.ietf.org https://www1.ietf.org/mailman/listinfo/mpls
>>>>>>
>>>>>> _______________________________________________
>>>>>> mpls mailing list
>>>>>> mpls@lists.ietf.org https://www1.ietf.org/mailman/listinfo/mpls
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> mpls mailing list
>>>>> mpls@lists.ietf.org https://www1.ietf.org/mailman/listinfo/mpls
>>>>
>>>
>>> _______________________________________________
>>> mpls mailing list
>>> mpls@lists.ietf.org https://www1.ietf.org/mailman/listinfo/mpls
>>
>> _______________________________________________
>> mpls mailing list
>> mpls@lists.ietf.org
>> https://www1.ietf.org/mailman/listinfo/mpls
>>

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