The MPLS WG Archive

["Adrian Farrel" <adrian@olddog.co.uk>]

Hi Neil,

Speaking as a co-author of draft-ietf-mpls-p2mp-sig-requirement-03.txt and
of draft-yasukawa-mpls-p2mp-oam-reqs-00.txt I can give an answer. Seisho
may want to comment further.

"protected by" is certainly not supposed to suggest that each and every
fault can be rectified by fast error recovery mechanisms. And, of course,
it would be crazy to expect that since failures of, for example, the
ingress router are not recoverable.

So, yes, "protected by" refers to defects that are protectable and
reported/detected, and only applies if the service has actually been
protected.

As you say, fault rectification is only going to take place if it is
detected in or reported to the layer in which we are deploying P2MP MPLS
TE. Further, as we attempt to say in the P2MP OAM requirements draft
(and/or inherit from the P2P OAM requirements draft) the higher layer must
be careful not to react too precipitously to a fault which the lower layer
may be in the process of solving.

With regard to "non-objectives". The non-objectives stated are NOT
non-objectives of the MPLS WG wrt to P2MP TE signaling. They are
non-objectives of *this* I-D. As the specific OAM subsection states, P2MP
OAM is discussed in a separate I-D.

There are three reasons for this:
1. OAM is not really a signaling requirement.
2. It is sufficiently large and important that folding it into the
signaling requirements I-D would make the signaling requirements I-D
unwieldy.
3. P2MP OAM needs to apply more widely than just TE (e.g. to LDP
multicast)

Cheers,
Adrian

----- Original Message ----- 
From: <neil.2.harrison@bt.com>
To: <yasukawa.seisho@lab.ntt.co.jp>
Cc: <mpls@ietf.org>
Sent: Wednesday, July 06, 2005 11:32 AM
Subject: RE: [mpls] I-D ACTION:draft-ietf-mpls-p2mp-sig-requirement-03.txt


Seisho,

I have a question on this draft:

In section 1 there is the following stated requirement:
"A P2MP TE LSP will be protected by fast error recovery mechanisms to
minimize disconnection of a P2MP service."

Which is obviously sensible, and of course implies that all the defect
cases of interest can be detected and the appropriate consequent actions
taken.  Note this should include defects of this layer network and not
just defects that arise in lower layer networks.  And remember....the
links between nodes in this layer network are provided by (p2p) trails
in a lower layer network....and this client/server relationship recurses
to the duct.  The point I am making here is that IF the lower layer
network has decent OAM itself you should get some indication of failure
from this, but this is over/above any defects you detect in this layer
network either from (i) lower layer defects or (ii) defects arising in
this layer network (which of course cannot be detected at all in any
lower layer network).


Yet it says in section 1.1 that a 'non-objective' is:
"OAM for P2MP LSPs"

Can someone please explain this apparent contradiction?


regards, Neil

_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls



_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls