[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Label type to be used

At 01:16 PM 3/21/2004 -0500, Kireeti Kompella wrote:

On Sun, 21 Mar 2004, dimitri papadimitriou wrote:

Hi Dimitri,

> the proposal as it read is:
>  >       [TDM, TDM] - label represents a TDM time slot [GMPLS-SONET-SDH]
>  >       [PSC, TDM] - fully transparent signal: label represents a port
>  >                    ("transparency" is defined in [GMPLS-SONET-SDH])
>  >       [PSC, TDM] - non-transparent signal: label represents a TDM time
>  >                    slot [GMPLS-SONET-SDH]
>
> the suggestion is also indicate the distinction for [TDM,TDM]:
>          [TDM, TDM] - fully transparent signal: label represents a port
>                       ("transparency" is defined in [GMPLS-SONET-SDH])
>          [TDM, TDM] - non-transparent signal: label represents a TDM time
>                       slot [GMPLS-SONET-SDH]
>  >       [PSC, TDM] - fully transparent signal: label represents a port
>  >                    ("transparency" is defined in [GMPLS-SONET-SDH])
>  >       [PSC, TDM] - non-transparent signal: label represents a TDM time
>  >                    slot [GMPLS-SONET-SDH]

Okay. Makes sense. Questions for all:

1) do you agree with this change?
2) if you implement GMPLS for [TDM, TDM] links, do you use port labels
   or SUKLM labels for the fully transparent case?
   - if SUKLM, would it be a hardship to change this?

Please respond by Sun Mar 28th, 5pm PST.

Thanks!
Kireeti.
-------

Kireeti,
I think Dimitri's clarification reflects the current state of the documents. So is *not* a change, but makes more explicit what is already in the documents. There are implementations that do exactly this today.

Changing to SUKLM, per your item 2, implies an update to [GMPLS-SONET-SDH]. Also, more importantly, (at least to me) is it implies that a link's switching type or label type may need change based on a neighbor's switching type! Consider transport equipment that support SONET/SDH framing, but not sub-channel switching. Item 2 implies that a SUKLM label would be used when the neighbor is a TDM switch, and a port label is used when a neighbor is a router. This seems wrong to me.

An alternative is to advertise such interfaces as LSC all the time. I don't think such advertisements are implied today, but the text could be change. An even better alternative is to introduce a new switching capability between TDM and LSC that is used for interfaces that only support a single transparent service on the interface. I like this alternative, but it is a bit late to make such a change...

Lou