Re: [CCAMP] Generalized Labels for the Flexi-Grid in LSC Label Switching Routers

["Giovanni Martinelli (giomarti)" <[e-mail]>]

Hi Adrian,

On 03 Mar 2014, at 18:26, Adrian Farrel <[e-mail]> wrote:

> Hey Giovani,
> 
>>> The difference is, in OTN the data plane can support hitless adjustment of
> the
>> bandwidth, thus the need of explanation of interaction of both planes; not
> sure it
>> is the case for flexi grid at the moment.
>> 
>> GM> IMHO yes.
> 
> I think you are saying that in your opinion hitless resizing in flexigrid is
> possible.
> 
> We might need to be careful with terminology. Some people say "hitless" and
> allow a few frames to be dropped. Some mean no frames will be dropped.
> 

GM> resizing as  “adding more bandwidth” to an existing LSP without hit on traffic alreday running on the LSP itself.  Now, as you know I’m an old-optical guys so in prinpriple I don’t know the traffic on top of the SSON LSP (either otn, packet etc.). For sure resizing in “narrowing the bandwidth” of existing LSP should be managed with care.
 

> We asked Q6/15, if I recall correctly, and got told "no" (need to check the
> liaison?). That does not preclude vendors making products that support this.
> 

GM> undestood. However sounds to me a similar problem to spectrum granularity … sometime technlogy is a bit ahead of std bodies.

> From a GMPLS perspective, I think we do support hitless reassignment of labels.
> It is as complicated as it would be if you used the management plane because
> coordination is needed between all network nodes.

GM> agree that we do something with existing GMPLS.  While writing not sure what the best to do, either leaving everything as is or at least spending few words (may be in fwk document?) about implications of having a certain label format. May be some flex grid fas has other ideas


> It is far nicer to do
> make-before-break onto a completely different piece of spectrum.
> 

GM> well, this sounds to me more like an operational choice… btw in this case I would say likely not hitless

Cheers,
G


> Cheers,
> Adrian
>