The MPLS WG Archive

[Thomas Nadeau <tnadeau@cisco.com>]

> I
> meant painful from the point of view of the number of places it could
> appear in the MIBs (which might not actually be that bad and looks like
> it could be confined to the XCTable).  I'd assume the most useful
> definition would be configurable per LSP with a default LSR setting. (per
> LSP default is default specified for the LSR).

        

Appearing
in more than one place is not such a bad thing
if it is useful, and not done too often.

        

--Tom

 
cheers
Dave.

-----Original Message-----
From: Thomas Nadeau [mailto:tnadeau@cisco.com]
Sent: Monday, November 26, 2001 3:00 PM
To: Allan, David [CAR:NS00:EXCH]
Cc: mpls
Subject: RE: MIBs and TTL

Pipe or uniform is the terminology adopted in the MPLS diffserv draft and carried forward into
http://www.ietf.org/internet-drafts/draft-agarwal-mpls-ttl-01.txt

        Sorry, from the original context, I wasn't sure.

The short story is:
- for uniform model TTL is independent of MPLS level. You copy TTL values up and down the stack as you push and pop.
- for pipe model, TTL is unique to each MPLS level.
So configuring this looks painful (get it from incoming label before x-connect entry, or get it from incoming IP payload, or set it to fixed value at ingress) :-(.

        It looks painful, but is flexible as to support different implementations.

As to how to handle non-TTL capable links, an unsigned object added to InterfaceConfEntry (InterfaceTTLDec?) with a default value of 1 should be all that is required. The only wrinkle being that I don't fully grok pop and swap to be sure that a non-ttl link label under the top label would work correctly.

        This is certainly an option.

        --Tom
        



------------------------------------------------------------------------
Mathematics is the supreme nostalgia of our time.

------------------------------------------------------------------------

Mathematics is the supreme nostalgia of our time.