The MPLS WG Archive

[Yakov Rekhter <yakov@juniper.net>]

Shahram,

> > While we are busy arguing about whether MPLS is a separate 
> > layer or not,
> > please bear in mind that the whole concept of a layer assumes 
> > (among other
> > things) a single format for transport of the data from the 
> > layer above.
> > Since MPLS doesn't make such an assumption (as in some cases it uses
> > the "shim" header, in other cases it uses the VCI/VPI, 
> > etc...) one could
> > conclude that perhaps MPLS doesn't fit into the traditional layering
> > model at all.
> 
> It seems that you have confused the "layer above" with the "layer below".
> X over MPLS always requires a Shim header. 

First of all, in the case of ATM-LSRs the presence of the shim header
strictly speaking is not mandatory. Second, even when the shim header
is present with ATM-LSRs, the forwarding information used by ATM-LSRs is 
*not* in the shim header, but in VCI/VPI fields of the ATM cell header. 
Finally, in GMPLS there is no shim header at all.

Yakov.