Re: [CCAMP] R: R: G.709 signaling - encoding Type

[Lou Berger <[e-mail]>]

Sergio,
	Thank you for clarifying why you/Daniele raised this topic.  My
preference/recommendation would be to not make any change.  I have
multiple reasons for this, primarily:

1) TSG is already unambiguous, although it is true that it must be
   derived.  Adding a second field that carries it explicit allows for
   inconsistencies and therefore adds error cases that must be detected
   and handled.

2) I'm not a big fan of guessing what the future holds.  I do agree
   that we can't preclude what might happen, and need to be supported,
   but I don't see a lot of value in defining mechanisms today that
   aren't needed today and might just be useful in the future.  In this
   particular example, if this ever really becomes an issue, we can
   just define a new label format.

3) We've had this discussion at least once before and agreed to the
   current approach.  Revising past decisions is fine if there's good
   cause, such as new information or issues identified/discovered, but
   this doesn't seem to apply in this case.

This said, I think we all would prefer to not prematurely limit
discussion on this, but also don't want to drag this on too long.

So it's really important that if there is anyone who supports making a
change, that they should speak up and state their justification for the
change (and soon)...

Thanks,
Lou

On 3/20/2013 10:04 AM, BELOTTI, SERGIO (SERGIO) wrote:
> Hi Lou,Fatai
> 
> In my summary you are reffering to I said:
> 
> ">> On the other hand this discussion is not new , since many time we discussed the opportunity to have dedicated object/fields to explicit adaptation information , at the moment not present in GMPLS."
> 
> It is recognized by everybody that we can retrieve information
> regarding mapping by the triple you indicated. The discussion is if
> it is the case to consider the opportunity to make explicit (in the
> label? ) the adaptation information . The fact that now with TSG we
> are able to understand the present mapping adopted in OTN, cannot
> assure in the future that for any possible extension regarding
> adaptation this triple can not be sufficient.
> 
> Since I'm one of the author but I'm not alone ..I cannot decide for
> everybody but just warn about.
> 

> 
> Best Regards
> 
> Sergio 
> 
> Belotti Sergio-  System Architect
> ALCATE-LUCENT  Optics Division
> via Trento 30 Vimercate (MB) - Italy
> phone [phone]
> 
> -----Messaggio originale-----
> Da: Lou Berger [mailto:[e-mail]] 
> Inviato: mercoledì 20 marzo 2013 14.53
> A: Fatai Zhang
> Cc: BELOTTI, SERGIO (SERGIO); CCAMP ([e-mail])
> Oggetto: Re: [CCAMP] R: G.709 signaling - encoding Type
> 
> Great.  Will you propose/add some related text so we don't need to
> revisit this again?
> 
> Lou
> 
> On 3/20/2013 5:27 AM, Fatai Zhang wrote:
>> Hi Lou,
>>
>> I agree on your conclusion, which I have pointed out one or two years ago when we discussed if it needs mapping information.
>>
>> I think some people may forgot this. 
>> =========================================================================================
>> Per the presented slide and Table 7-10 of G.709, the mapping method/type is unique per combination of {ODUj, ODUk, TSG}.
>>
>>
>>
>>
>> Best Regards
>>
>> Fatai
>>
>>
>> -----Original Message-----
>> From: [e-mail] [mailto:[e-mail]] On Behalf Of Lou Berger
>> Sent: Friday, March 15, 2013 7:02 PM
>> To: BELOTTI, SERGIO (SERGIO)
>> Cc: CCAMP ([e-mail])
>> Subject: Re: [CCAMP] R: G.709 signaling - encoding Type
>>
>> Sergio,
>> 	Thank you for clarifying the requirement. I believe you are describing
>> an issue that is a little different than what I understood from some
>> earlier discussions.  Perhaps others were also in the same situation.
>>
>> As I now understand it, the issue that is being raised is one related to
>> ODUj LSPs that use an ODUk H-LSP as a hop, there is a need at all hops,
>> including at *transit nodes*, to configure the mapping method (AMP, GMP)
>> used from the ODUj onto the ODUk H-LSP.  This of course means that the
>> mapping function needs to be set on a hop by hop basis for the ODUj LSP.
>>  (I though an issue on and ODUk H-LSP was being raised.)
>>
>> Or in other words there is an issue of per-hop resource
>> allocation/management.
>>
>> Per the presented slide and Table 7-10 of G.709, the mapping method/type
>> is unique per combination of {ODUj, ODUk, TSG}.  Obviously any hop
>> signaling an ODUj LSP will know and unambiguously be able to identify
>> (in signaling) the ODUj and ODUk.  This just leaves the TSG.
>>
>> Section 6.1 of the signalling draft already has text covering TSG as
>> part of the per-hop label. So TSG is also already being signaled
>> unambiguously.
>>
>> It looks to me that all that is needed is a sentence or two stating that
>> the mapping type is unambiguously based on the above.
>>
>> Please let me know if I missed something.
>>
>> Lou
>>
>> On 3/15/2013 6:05 AM, BELOTTI, SERGIO (SERGIO) wrote:
>>>
>>> The requirement comes form discussion had last week related to tolerance .
>>> We discovered from ITU experts, there is an attribute related to the adaptation function in case of multiplexing of ODUj into ODUk, 
>>>
>>>  oduTypeAndRate that configures the mapping method (AMP, GMP).
>>>
>>> Now , since obviously equipment cannot know whether configuration is provided via management or control plane , conclusion was that control plane should have an equivalent of this attribute in order to be able to configure the right adaptation.
>>>
>>> On the other hand this discussion is not new , since many time we discussed the opportunity to have dedicated object/fields to explicit adaptation information , at the moment not present in GMPLS.
>>>
>>> This is a summery of requirement.
>>>
>>> Thanks
>>> Sergio
>>>
>>> Belotti Sergio-  System Architect
>>> ALCATE-LUCENT  Optics Division
>>> via Trento 30 Vimercate (MB) - Italy
>>> phone [phone]
>>> -----Messaggio originale-----
>>> Da: Lou Berger [mailto:[e-mail]] 
>>> Inviato: giovedì 14 marzo 2013 19.58
>>> A: Daniele Ceccarelli
>>> Cc: Rajan Rao; John E Drake; BELOTTI, SERGIO (SERGIO); CCAMP ([e-mail])
>>> Oggetto: Re: [CCAMP] G.709 signaling - encoding Type
>>>
>>> Daniele, (Sergio?)
>>>
>>> Can you summarize the data plane behavior/requirement that is the basis
>>> for the change?  i.e., provide the reason why *any* change is required.
>>>
>>> Once we all understand the data plane requirements/constraints, we can
>>> better decide which (hopefully existing) GMPLS mechanism is best suited
>>> to support the requirement.
>>>
>>> Thank you,
>>> Lou
>>>
>>> On 3/14/2013 2:41 PM, Daniele Ceccarelli wrote:
>>>> Using the encoding was one of the possible suggestions.
>>>>
>>>> Indeed it has impacts on the routing and it might be worth avoiding it, but another solution is preferrable wrt relying on the Label Length (which might change for several reasons). I would prefer not to put a requirement on the Label lenght only because it is needed to retrieve the adaptation type.
>>>>
>>>> Other proposals are welcome. I had a chat with John this morning and he was proposing the utilization of a new field (even a single bit). That could be a viale option for me. Since we're removing the Tolerance from the traffic parameters we're going to have room for a new field.
>>>>
>>>> BR
>>>> Daniele
>>>>
>>>>> -----Original Message-----
>>>>> From: Rajan Rao [mailto:[e-mail]] 
>>>>> Sent: mercoledì 13 marzo 2013 19.51
>>>>> To: Lou Berger; John E Drake
>>>>> Cc: BELOTTI, SERGIO (SERGIO); CCAMP ([e-mail]); Daniele 
>>>>> Ceccarelli
>>>>> Subject: RE: [CCAMP] G.709 signaling - encoding Type
>>>>>
>>>>> Lou,
>>>>>
>>>>> The new encoding has implications in routing(new ISCDs).  I 
>>>>> think we can minimize the impact using the two fields 
>>>>> mentioned in my email below.   To be specific,
>>>>>
>>>>> For the case highlighted in the slide,  the Encoding is AMP 
>>>>> when Signal Type = ODU1  in traffic spec Length Field  = 8 in  Label 
>>>>>
>>>>> Thanks
>>>>> Rajan
>>>>> -----Original Message-----
>>>>> From: Lou Berger [mailto:[e-mail]]
>>>>> Sent: Wednesday, March 13, 2013 6:54 PM
>>>>> To: John E Drake
>>>>> Cc: BELOTTI, SERGIO (SERGIO); Rajan Rao; CCAMP 
>>>>> ([e-mail]); Daniele Ceccarelli
>>>>> Subject: Re: [CCAMP] G.709 signaling - encoding Type
>>>>>
>>>>> John,
>>>>>
>>>>> See below.
>>>>>
>>>>> On 3/13/2013 6:40 PM, John E Drake wrote:
>>>>>> Lou,
>>>>>>
>>>>>>  
>>>>>>
>>>>>> I must have been asleep but I don't remember hearing of an issue.
>>>>>
>>>>> No problem, looked like a few others had some trouble getting 
>>>>> moving this AM too.
>>>>>
>>>>>>  It
>>>>>> was my understanding that AMP and GMP both use G.709 encoding in the 
>>>>>> data plane, so why would we want to make what appears to be an 
>>>>>> artificial distinction?
>>>>>
>>>>> This was covered on slide 3 of Danielle's presentation.  
>>>>> He/They can provide additional details/justification.
>>>>>
>>>>> Lou
>>>>>
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Irrespectively Yours,
>>>>>>
>>>>>>  
>>>>>>
>>>>>> John
>>>>>>
>>>>>>  
>>>>>>
>>>>>> *From:*Lou Berger [mailto:[e-mail]]
>>>>>> *Sent:* Wednesday, March 13, 2013 3:27 PM
>>>>>> *To:* John E Drake; BELOTTI, SERGIO (SERGIO); Rajan Rao
>>>>>> *Cc:* CCAMP ([e-mail]); Daniele Ceccarelli
>>>>>> *Subject:* Re: [CCAMP] G.709 signaling - encoding Type
>>>>>>
>>>>>>  
>>>>>>
>>>>>> John,
>>>>>>
>>>>>> Do you have an alternate proposal on how to address the issue, or do 
>>>>>> you just see an issue?
>>>>>>
>>>>>> (If the former, the onus will fall on you to provide one. If the 
>>>>>> latter, it'll fall to Sergio And Danielle to recap the presented
>>>>>> issue.)
>>>>>>
>>>>>> Lou
>>>>>>
>>>>>>  
>>>>>>
>>>>>> On March 13, 2013 6:16:46 PM John E Drake wrote:
>>>>>>
>>>>>>     Hi,
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     I don't think this is a good idea and I don't see any 
>>>>> reason for it.
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     Irrespectively Yours,
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     John
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     *From:*[e-mail] <mailto:[e-mail]>
>>>>>>     [mailto:[e-mail]] *On Behalf Of *BELOTTI, 
>>>>> SERGIO (SERGIO)
>>>>>>     *Sent:* Wednesday, March 13, 2013 6:53 AM
>>>>>>     *To:* Rajan Rao
>>>>>>     *Cc:* CCAMP ([e-mail] <mailto:[e-mail]>)
>>>>>>     *Subject:* [CCAMP] R: G.709 signaling - encoding Type
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     Rao,
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     it is a proposal: so you should read:  "encoding type can 
>>>>>> indicate.
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     Regards
>>>>>>
>>>>>>     Sergio
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     *Belotti Sergio-  System Architect*
>>>>>>
>>>>>>     *ALCATE-LUCENT  Optics Division*
>>>>>>
>>>>>>     via Trento 30 Vimercate (MB) - Italy
>>>>>>
>>>>>>     phone [phone]
>>>>>>
>>>>>>     
>>>>>>
>>>>> ----------------------------------------------------------------------
>>>>>> --
>>>>>>
>>>>>>     *Da:*[e-mail] <mailto:[e-mail]>
>>>>>>     [mailto:[e-mail]] *Per conto di *Rajan Rao
>>>>>>     *Inviato:* mercoledì 13 marzo 2013 14.40
>>>>>>     *A:* CCAMP ([e-mail] <mailto:[e-mail]>)
>>>>>>     *Oggetto:* [CCAMP] G.709 signaling - encoding Type
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     Daniele,
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     The presentation slide#3 says:  "encoding type indicates AMP or
>>>>>>     GMP".   I don't think this is the case.  We use  G.709 ODUk as
>>>>>>     encoding type.  There is no explicit indication of AMP or GMP
>>>>>>     there.  Are you proposing to change this?
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     Note that AMP/GMP can be inferred from Signal Type in 
>>>>> traffic param
>>>>>>     & Length field ( = 8) in the label.
>>>>>>
>>>>>>      
>>>>>>
>>>>>>     Thanks
>>>>>>     Rajan
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>>
>> _______________________________________________
>> CCAMP mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/ccamp
>>
>>
>>
>>
> 
> 
> 
>