The MPLS WG Archive

["Andrew G. Malis" <Andy.Malis@vivacenetworks.com>]

Dan,

It depends on the usage - if you're using FR switches as LSRs, and using 
MPLS as the control plane to allocate the DLCIs, then you must use RFC 
3034.  However, if you're simply tunneling MPLS packets through a FR 
network which uses a native control plane, or perhaps FRF.5 across an ATM 
backbone and control plane, then you are correct.

I was aware of a couple of implementations of RFC 3034 at the time we were 
working on the draft.  Unfortunately, one of the places has closed up shop, 
and the other seems to have been either discontinued or never made it into 
a shipping release.  It looks like the RFC may never make it past Proposed ....

Cheers,
Andy

-------

At 1/6/2003 09:37 AM -0500, Dan Tappan wrote:

>At 10:20 AM 1/5/2003 -0500, Andrew G. Malis wrote:
>>http://www.ietf.org/rfc/rfc3034.txt is the reference.  See section 4 for 
>>the encapsulation.  As you'll see, it's very straightforward.
>>
>>In particular, no NLPID (for multiprotocol identification) is required 
>>because FR VCs used as MPLS LSPs are dedicated to that use.  That's why 
>>you couldn't find a specific reference.
>
>While the above is technically correct, I don't know of anyone who has 
>actually implemented rfc3034.
>
>A more common approach for carrying MPLS across a FR link is to follow 
>rfc2427: using the LLC/SNAP (section 4.1) encapsulation above the rfc3032 
>encapsulation.
>
>>Cheers,
>>Andy