The MPLS WG Archive

[Curtis Villamizar <curtis@workhorse.fictitious.org>]

In message <E1AM2Vd-00023L-00@celiborn.ip-engineering.bt.com>, Peter Willis wri
tes:
> If we don't document OAM issues with MPLS in the OAM framework document where
>  
> do they get documented? 
> 
> I think the precedent for discussing Issues with MPLS in MPLS WG framework
>  documents is established in RFC3353.
> 
> Regards the "*practical*" point I can see why the MPLS WG has problems with 
> the top down nature of draft-allan-mpls-oam-frmwk-05.txt. Perhaps the usual 
> modus operandi of documenting implementations and waiting for customers to
> request features to fix operational problems works much better?
> 
> Peter.


An informational RFC on P2MP has nothing to do with this.  There was
very limited interest in P2MP and no one considered that framework to
be harmful.  If good concrete work comes of it that would be welcome.

If there are aspects of the architecture that impacts some methods of
OAM (and clearly there is), it should be covered.  Neither the method
of OAM should be called deficient, nor the architecture, it just
becomes an applicability issue.

There are some people who see the need for monitoring in networks that
have ECMP and have no intention of removing ECMP.  These are real
networks and therefore this is a real problem.  Some methods of
monitoring will work others won't.  That is an issue of applicability.

Some people (you, Neil, Dave, others) prefer aspects of one approach.
Some people prefer the broader applicability of other approaches.
These are tradeoffs.  If we accept these as tradeoffs, then we can
move forward.  If some people keep insisting that we need a document
that states that the architecture is broken we can make no progress.

Curtis