RE: [PWOT] mpls-in-ip

From: Sasha Vainshtein [mailto:Sasha@AXERRA.com]
>
> 	I have doubts regarding your presumtions:
> 	1)	An MPLS LSP may be multipoint-to-point as 
> opposed to all the
> rest of of the pseudo-wire examples in 		your message

in mpls-in-ip no label switching occurs between 
encapsulator and decapsulator. so i don't see
that p2mp and mp2p lsr requirements affect the
encapsulation.

> 	2)	IP per se is can be hardly considered as 
> transport since it
> does not allow for demuxing between an
> 		indefinitely large number of logical 
> application instances
> within the downstream host. BTW, this is why 		- IMHO 
> - in order to
> treat MPLS as transport you usually require a stack of two 
> labels while ATM
> provides transport capabilities inherently by using VPI/VCI

if such multiplexing is, in the opinion of the wg, a
necessary characteristic of "transport" then i would
argue that mpls-in-ip is a good example of signalled 
multi-protocol transport tunnels for ip core networks.

c u
fsb

_______________________________________________
pwot mailing list
pwot@ietf.org
http://www.ietf.org/mailman/listinfo/pwot