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

Javi wrote:
> Ignacio,
> 
> RFC 3931 (L2TPv3), clause 5.4.4
> Note that a Session ID value is *unidirectional*.  Because each LCCE
> chooses its Session ID independent of its peer LCCE, the value
> does not have to match *in each direction* for a given session.
> 
> I understand each session is unidirectional, and two sessions (one for 
> each direction) are needed for bidirectional traffic.

No, an L2TP session is always 100% bidirectional: traffic can flow
in both directions.

It is only the IDs used to identify incoming packets at each end that
happen to be independent.

In other words, if you have an L2TP session opened between LCCE1
and LCCE2, traffic going from LCCE1 to LCCE2 will have a Session ID
that is independent (and potentially different) from the Session
ID used on traffic going from LCCE2 to LCCE1.

Although both directions have different IDs, they are halves of
the _same_ L2TP session. L2TP does not have the capability to create
unidirectional sessions.

If you like to think in physical terms, imagine a cable (session)
with a red connector in one end and a black connector in the other.
Although the ends have different colors, they are still ends of the
same cable. Any signal you send from either end will show up at
the other end of the cable, regardless of the connector color.


> A Session ID is a possible PW label. So, a L2TPv3 session is really a 
> L2TPv3 PW.

You are confusing terms and concepts.
Don't confuse labels with sessions with traffic within a session.
They are all very different concepts.

-Ignacio