Re: [mpls-tp] MPLS-TP tunnel and PW establishment over multipledifferent operators

[<[e-mail]>]

Here is my view of some of the issues that should be
understood/considered here:

-	the most important use-case of a transport network is creating
client layer network topology.  Now the client can also be the same
technology as the server...and here being able to provide
MPLS-TPoverMPLS-TP as a proper client/server relationship will be key to
the success of MPLS-TP as a transport network.  In a client/server
relationship the DP/CP/MP of the client and server layer networks are
fully functionally isolated/independent...so the issue you raise below
(and which only considers DP addressing mismatching) is a non-issue ina
proper client/server case.

-	you are showing a E-NNI relationship, ie a single layer network
that is peered between 2 different parties.  This implies a horizontal
partitioning of a single layer network....and so we have to consider the
interoperability of *all* the DP/CP/MP components, ie it is more than
simple DP access point addressing.  Note carefully that when components
are not idetical then we have a protocol/function converison problem to
deal with...this case may have no solution at all if the 2 technolgies
do not belong to same network mode (eg one is cl-ps and the other co-ps
or co-cs) and there is oftem significant misalignment when the 2
technologies belong to the same mode.

-	also note carefully that when peer interworking we do not nor
normally peer the MP, and we restrict the scope of the CP signalling and
routing components....in other words an E-NNI between nodes belonging to
different parties is very different to an I-NNI between nodes belonging
to the same party.

-	all the above is all academic in any case for MPLS-TP as we
should never peer network technologies that are not TOS
(Top-Of-Stack)....most technologies are NOT TOS.  A TOS network directly
supports external message/file/stream applications via key application
adaptation functions....very few network technologies have these, eg IP
has (UDP/TCP/RTP), ATM has (AALs) and the PSTN has (eg various voice
(stream) encoding schemes such as A/u-law PCM sampling quantizing).  A
TOS layer network exists in *all* networking scenarios...if it does not
then all lower layer networking serves no purpose at all.  By definition
a non-TOS network MUST carry a TOS layer network.  It therefore
immediately follows that there is no point whatsoever in peer
interworking a non-TOS layer network across an E-NNI, we should simply
fully terminate the DP/CP/MP of the non-TOS layer network at the
interworking point and pass across the higher layer TOS layer network
transparently.

Aside=> One see this error of pushing unnecessary E-NNIs in all sorts of
technology promoting fora...I suppose its politically/commercailly
understandable ('rule the world' minset), even if obviously technically
unnecessary wrt the work and costs/problems it creates.

Hope that helps.

regards, Neil. 

> -----Original Message-----
> From: [e-mail] 
> [mailto:[e-mail]] On Behalf Of venkatesan mahalingam
> Sent: 01 July 2010 08:54
> To: [e-mail]
> Subject: Re: [mpls-tp] MPLS-TP tunnel and PW establishment 
> over multipledifferent operators
> 
> Hi,
> 
> I guess transport people will be able to provide more 
> pointers and suggestions against such deployment 
> possibilities for MPLS_TP.
> Is this not a valid scenario that needs to be discussed?
> 
> Thanks,
> Venkat.
> 
> On 6/30/10, venkatesan mahalingam <[e-mail]> wrote:
> > Hi,
> > Can transport guys please reply to the below query?
> >
> > Thanks,
> > Venkat.
> >
> > On 6/30/10, venkatesan mahalingam <[e-mail]> wrote:
> >> Hi,
> >> Can we establish a LSP/PW between MPLS-TP operators with different 
> >> identifiers (ICC and Global_ID)?
> >>
> >> for example,
> >>
> >> Operator1 
> ----------Operator2----------------Operator3----------------Operator-4
> >>  (ICC)                    (ICC)                    
> (Global_ID)              (Global_ID)
> >>
> >> Can single LSP/PW be traversed between Operator1 and Operator-4?
> >> If the answer is NO, do we need to maintain two LSPs/PWs, 
> one between
> >> Operator1 and Operator2 and
> >> another one between Operator3 and Operator4 and LSP/PW 
> stitching with 
> >> these two LSPs/PWs for communication across operators?
> >>
> >>
> >> Thanks,
> >> Venkat.
> >>
> >> On Mon, Jun 28, 2010 at 11:13 PM, venkatesan mahalingam < 
> >> [e-mail]> wrote:
> >>
> >>> Hi,
> >>>
> >>> Please clarify whether the below operations are valid for MPLS-TP 
> >>> deployment.
> >>>
> >>> For example,
> >>> *Scenario-1:*
> >>> Operator1 
> -------------------Operator2--------------------Operator3
> >>>  (ICC - IP capable)    (ICC- IP capable)        (ICC IP incapable)
> >>>
> >>>  Can single LSP/PW be established over ICC based 
> operators in an IP 
> >>> and Non-IP environments?
> >>>  **
> >>> *Scenario-2:*
> >>> Operator1 
> -----------------Operator2--------------Operator3-------------
> Operator-4
> >>>  (ICC - IP capable)   (ICC- IP capable)   (ICC IP 
> incapable)      (ICC IP incapable)
> >>>
> >>> Can single LSP/PW be established between Operator1 and 
> Opearator-4?
> >>> If the answer is NO, do we need to maintain two LSPs/PWs, one 
> >>> between
> >>> Operator1 and Operator2 and
> >>> another one between Operator3 and Operator4 and LSP/PW stitching 
> >>> with these two LSPs/PWs?
> >>>
> >>>  *Scenario-3:*
> >>> Operator1
> >>> 
> ---------------------------Operator2--------------------------
> ------Operator3
> >>> (Global_ID - IP capable)    (Global_ID - IP capable)      
>   (Global_ID
> >>> IP
> >>> incapable)
> >>>
> >>>  Can single LSP/PW be established over Global_ID based 
> operators in 
> >>> an IP and Non-IP environments?
> >>>  **
> >>> *Scenario-4:*
> >>> Operator1
> >>> 
> -----------------------------Operator2------------------------
> Operator3------------------------------Operator-4
> >>>  (Global_ID - IP capable)    (Global_ID- IP capable)      
>   (Global_ID
> >>> IP
> >>> incapable)       (Global_ID IP incapable)
> >>>
> >>> Can single LSP/PW be established between Operator1 and 
> Opearator-4?
> >>> If the answer is NO, do we need to maintain two LSPs/PWs, one 
> >>> between
> >>> Operator1 and Operator2 and
> >>> another one between Operator3 and Operator4 and LSP/PW stitching 
> >>> with these two LSPs/PWs?
> >>>
> >>> *Scenario-5:*
> >>>  Operator1
> >>> 
> --------------------------------Operator2--------------------Operator3
> >>>  (Global_ID- IP capable)         (ICC- IP capable)        
> (Global_ID IP
> >>> capable)
> >>>
> >>>  Can single LSP/PW be established over different 
> operators in an IP 
> >>> environments?
> >>>  **
> >>>  *Scenario-6:*
> >>>  Operator1
> >>> 
> -----------------------------Operator2--------------------Operator3
> >>>  (ICC- IP incapable)         (Global- IP incapable)        (ICC IP
> >>> incapable)
> >>>
> >>>  Can single LSP/PW be established over different operators in an 
> >>> Non-IP environments?
> >>>
> >>> *Scenario-7:*
> >>> Operator1
> >>> 
> -------------------Operator2--------------------Operator3-----
> ----------------------Operator-4
> >>>  (ICC - IP capable)    (ICC- IP capable)        (Global_ID IP
> >>> capable)            (Global_ID IP capable)
> >>>
> >>> Can single LSP/PW be established between Operator1 and 
> Opearator-4?
> >>> If the answer is NO, do we need to maintain two LSPs/PWs, one 
> >>> between
> >>> Operator1 and Operator2 and
> >>> another one between Operator3 and Operator4 and LSP/PW stitching 
> >>> with these two LSPs/PWs?
> >>>
> >>>  *Scenario-8:*
> >>> Operator1
> >>> 
> -------------------Operator2--------------------Operator3-----
> ----------------------Operator-4
> >>>  (ICC - IP capable)    (ICC- IP capable)        (Global_ID IP
> >>> incapable)          (Global_ID IP incapable)
> >>>
> >>> Can single LSP/PW be established between Operator1 and 
> Opearator-4?
> >>> If the answer is NO, do we need to maintain two LSPs/PWs, one 
> >>> between
> >>> Operator1 and Operator2 and
> >>> another one between Operator3 and Operator4 and LSP/PW stitching 
> >>> with these two LSPs/PWs?
> >>>
> >>>  *Scenario-9:*
> >>> Operator1
> >>> 
> -------------------Operator2-----------------------------Opera
> tor3---------------------------Operator-4
> >>>  (ICC - IP capable)    (Global_ID- IP capable)        (ICC IP
> >>> incapable)          (Global_ID IP incapable)
> >>>
> >>> Can single LSP/PW be established between Operator1 and 
> Opearator-4?
> >>> If the answer is NO, do we need to maintain two LSPs/PWs, one 
> >>> between
> >>> Operator1 and Operator2 and
> >>> another one between Operator3 and Operator4 and LSP/PW stitching 
> >>> with these two LSPs/PWs?
> >>>
> >>>  *Scenario-9:*
> >>> Operator1
> >>> 
> -------------------------Operator2------------------------Oper
> ator3---------------------------Operator-4
> >>>  (ICC - IP capable)    (Global_ID- IP incapable)        
> (Global_ID IP
> >>> capable)          (ICC IP incapable)
> >>>
> >>> Can single LSP/PW be established between Operator1 and 
> Opearator-4?
> >>> Is this a valid scenario?
> >>>
> >>> --
> >>> Best Regards,
> >>> Venkatesan Mahalingam.
> >>>
> >>
> >>
> >>
> >> --
> >> Best Regards,
> >> Venkatesan Mahalingam.
> >>
> >
> >
> > --
> > Best Regards,
> > Venkatesan Mahalingam.
> >
> 
> 
> --
> Best Regards,
> Venkatesan Mahalingam.
> _______________________________________________
> mpls-tp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/mpls-tp
>