Re: Pseudo-wires: uni-directional o bi-directional?

El 22/06/2010 4:14, Ignacio Goyret escribió:
> But note I didn't use the term "PW label". That is a term typically
> used with MPLS and not all PWs are carried over MPLS.
>
> So, let's clarify some terms: what do you call a "PW label"?
> Because that term is not really used in L2TP terminology.


I don't agree. "PW label" is a general term for any PW (L2TPv3, MPLS, 
UDP/IP, Ethernet, ...)

See RFC 3985 "Pseudo Wire Emulation Edge-to-Edge (PWE3) Architecture", 
clause "5.4.1.  PWE3 over an IP PSN"

    PW Demultiplexing is provided by the PW label, which may take the
    form specified in a number of IETF protocols;  e.g., an MPLS label
    [MPLSIP], an L2TP session ID [RFC3931], or a UDP port number
    [RFC768].


>
>> Then, why is a L2TPv3 session not a "L2TPv3 PW"?, which would be the 
>> difference?
>
> Be careful using a term like "L2TPv3 PW". That term can have two
> potential meanings: either a PW built using L2TPv3 services or
> an L2TPv3 session carried over some sort of PW.
> Which do you mean?

I understand a "L2TPv3 PW" is a PW which "PW Demultiplexer layer" (or PW 
label) is based on a L2TPv3 Session ID, using the L2TPv3 control plane.

An L2TPv3 session carried over some sort of PW (like a MPLS PW) would be 
different. Over a PW you may carry on any network service, such as 
L2TPv3, but in that case, the L2TPv3 session would not be part of the PW.

>
> At any rate, the difference is a matter of layers.
>
> L2TPv3 is used to build PWs, but not the other way around.

Ok, I agree with this.

>
>> And, finally, is a PW bidirectional, like a L2TPv3 Session?
>
> That's a question what should be asked on the PWE3 WG.
> This WG (l2tpext) concerns itself only with L2TP.

I did, but I have not received any answer yet :-)

>
> -Ignacio
>

Thanks,

Javi