The MPLS WG Archive

[Alia Atlas <aatlas@avici.com>]

At 12:02 PM 4/23/2002 -0400, Eric Rosen wrote:
>Markus> I certainly would have expected option 2!
>
>Well, the problem with option 2  is that MPLS forwarding likes to obtain the
>data link  header as  a result  of the label  lookup.  So  I would  say that
>option 2 isn't really an option.

But when you remove the last label, it is then necessary to do some IP 
header processing such as TTL decrementing or, at the least, 
inheriting.  Granted, there are different models (uniform, pipe, 
short-pipe) for handling TTL, but several of the models clearly presume 
that the LSR popping the last label must handle the TTL.  To handle the 
TTL, the LSR must know what type of packet is underneath (IPv4 or IPv6 or...).

Could you clarify how you'd envision the appropriate behavior here if the 
MPLS forwarding can't look underneath the MPLS shim header, when it is 
popping the last one?

This isn't a limitation in the MPLS forwarding layer which I am familiar with.

Thanks,
Alia