[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
WG: Moving right along ... Switching Type
Dimitri,
thanks for the clarification.
From this information it seems first to me that the values for the switching
type are miss-leading, instead of "TDM" or "WDM" we should use "time slots"
and "wavelengths". But furthermore what additional information does the
switching type provide in addition to the signal type. It seems that the
switching type is always defined by the signal type. A ODU1 signal type has
a switching type of TDM/time slot and a OCh a switching type of
WDM/Wavelength.
Can you or someone else give an example where the switching type is not
related to the signal type (e.g. signal type ODU1 and switching type
WDM/wavelength)?
For the first example you say that the upper ODU1 (OTM1.4) are identified by
individual OCh IDs. I assume that the lower ODU1 are identified by the ODU2
OCh ID and the timeslot within the ODU2.
It looks strange to me, that I have to use different methods for the label
generation depending on the server layers. This results in the problems with
example 2. Combining several signals using TDM or WDM is no difference from
the basic signal flow. It is a technology choice that may change over time.
GMPLS is in my view for this case too much focused on technologies. I would
prefer a label that is independent of the server layer technology. This
would allow to change the server layer at intermediate points without
problems.
Regards
Juergen
> -----Ursprüngliche Nachricht-----
> Von: Dimitri Papadimitriou [mailto:dimitri.papadimitriou@alcatel.be]
> Gesendet: Dienstag, 6. November 2001 07:31
> An: Heiles Juergen
> Cc: ccamp@ops.ietf.org
> Betreff: Re: Moving right along ... Switching Type
>
>
> Juergen, see inline...
>
> Heiles Juergen wrote:
> >
> > Following the discussion on the Switching Type and looking
> on the proposed G.709 extensions for GMPLS I would like to
> get some feedback on the examples shown below.
> > How to handle the examples concerning label request and
> returned label is important for my understanding of GMPLS.
> > The examples are based on G.709 ODUs. A ODU can be
> transported via WDM systems or using TDM in a higher bit rate
> ODU. So we can have a mixture of both methods.
> >
> > Juergen
> >
> > Example 1:
> >
> > Two network elements A and B with ODU1 cross-connect are
> interconnected by two types of signals.
> > - a four channel WDM OTM-4.1 interface with 4 optical
> channels each carrying an OTU1/ODU1
> > - a single channel OTM-1.2 with one optical channel
> carrying a OTU2/ODU2 and 4 ODU1 multiplexed into the ODU2.
> >
> > ----------------- ----------------
> > NE A + + NE B
> > + +
> > ODU1 ----- + + ----- ODU1
> > 1 o---+ + + OTM-4.1 + + +---o 1
> > 2 o---+ w + + 4xOCh with OTU1/ODU1 + + W +---o 2
> > + D +--------------------------------+ D +
> > 3 o---+ M + + + + M +---o 3
> > 4 o---+ + + + + +---o 4
> > ----- + + -----
> > + +
> > ODU1 ----- + + ----- ODU1
> > 5 o---+ + + OTM-0.2 + + +---o 5
> > 6 o---+ T + + 1xOCh with OTU2/ODU2 + + T +---o 6
> > + D +--------------------------------+ D +
> > 7 o---+ M + + with 4 ODU1 + + M +---o 7
> > 8 o---+ + + + + +---o 8
> > ----- + + -----
> > + +
> > ----------------- -----------------
> >
> > Now setting up a ODU1 connection between A and B. 8 ODU1
> connections are available, 4 using WDM technology and 4 using
> TDM technology.
> > For the connection itself I don't care which of the 8 is
> used and I don't care about the specific technology.
> > How does the label request from NE a to NE B look like?
> > - the LSP encoding type is "G.709 digital path layer"
> > - the signal type is ODU1
> > - but what is the switching type "TDM" or "Lambda"? Both
> are available.
>
> Both entities aren't ODU1 switches ? i think you have indicate in the
> picture here "the interface multiplexing" capabilities not
> the switching
> ones.
>
> So I guess that these devices are both ODU1 switches. Therefore, if
> the switching type is used (notice that it is not necessary here since
> there is only one switching capability assigned to the "end of the
> link") the switching will be TDM
>
> The differences comes from the Encoding type in the first case
> (upper link) it will be G.709 OCh (WDM interface) in the second
> case G.709 ODUk (TDM interface)
>
> > Furthermore what kind of label is returned by NE B to NE A?
> > - for ODU 5 to 8 a "G.709 digital path layer label" with
> signal type ODU1 and a indication of the time slot within the ODU2
> > - but how about ODU1 to 4, a G.709 OCh signal type with the
> indication of the wavelength is not correct.
> > or is each ODU 1 to 4 seen as individual link?
>
> In such a case (for the OTM4.1) each of the OCh will have a
> dedicate ID
> that will be included within the request: setup of an ODU1 LSP
> - LSP encoding type: G.709 digital path layer
> - signal type: ODU1
> - switching type: "TDM" (not necessary)
> - If_Index: 1 (for instance to indicate the first optical channel)
>
> > Example 2:
> >
> > NE A and NE C that provide ODU1 cross-connections are
> interconnected via NE B that is not GMPLS aware and that
> performs a conversion between WDM and TDM multiplexing.
> > - between NE A and NE B a four channel WDM OTM-4.1 signal
> is used with 4 optical channels, each carrying a OTU1/ODU1.
> > - between NE B and NE C a single channel OTM-0.2 signal is
> used with one optical channel carrying a OTU2/ODU2 with 4
> ODU1 multiplexed into the ODU2.
> >
> > ------------- ---------------- -------------
> > NE A + + NE B + + NE C
> > + + ODU1 + +
> > ODU1 ----- + + ----- ----- + + ----- ODU1
> > 1 o--+ + + OTM-4.1 + + +--+ + + OTM-0.2 + + +--o 1
> > 2 o--+ w + + 4xOCh with + + W +--+ T + + 1xOCh with + + T +--o 2
> > + D +----------------+ D + + D +-----------------+ D +
> > 3 o--+ M + + OTU1/ODU1 + + M +--+ M + + OTU1/ODU1 + + M +--o 3
> > 4 o--+ + + + + +--+ + + with 4 ODU1 + + +--o 4
> > ----- + + ----- ----- + + -----
> > + + + +
> > ------------- ---------------- -------------
> >
> > Again what is the switching type of the label request from
> NE A to NE C (TDM or Lambda)?
> > And what is the returned label from NE C to NE A?
>
> Just to be sure in the above example i guess you ment for the second
> link 1x0Ch with OTU2/ODU2 with 4 ODU1 since having TDM
>
> Since device A, B and C are all ODU1 switches the switching will be
> always
> TDM (if this field is used cfr. the remark indicated here above).
> Remember
> here that TDM is a shortcut to specify switching of all the digital
> switching
> layers supporting TDM capabilities (this includes VC-3/VC-4/ODUk/etc.)
>
> The relative complexity in the second example shown here above arise
> more from
> the interconnection of GMPLS to non-GMPLS capable devices (in addition
> performing
> conversion of the encoding type) rather than the "switching
> type" issue.
>
> It reminds me Zhi's discussion on GMPLS to non-GMPLS
> communication here
> addind that your transmission competences it gives a very
> nice example.
>
> Briefly LSR A sees ODU mapping while C sees ODU multiplexing such that
> building the label association is needed as additional
> mechanisms here.
>
> - dimitri.
>