[e2e] What should e2e protocols know about lower layers?

[Joe Touch]

Anoop Ghanwani wrote:

> I agree with Ran's analysis.  Especially with things
> like L2 VPNs over MPLS, there's no way to know how
> far the traffic will be going, or what kind of path
> it would take (routed, switched).


I hadn't addressed VPNs before; it seems appropriate to differentiate 
between "local subnet on physically-connected interface" from other 
subnets, including both remote subnets and subnets attached to virtual 
interfaces. Toss in things that are either inherently nonlocal or 
non-broadcast, such as ATM and PPP, and that might be sufficient.

Joe