[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: FW: path-decoupled signaling (pads) BOF
Cedric,
Sorry for the late reply. Please find comments inline.
-- Cedric Aoun wrote on 06 March 2003 10:15 +0100:
Martin and Juergen,
As the 3 of us spent time on Middlebox control we know the pain of off-path
signalling (topology awareness part) but we also know that it is better than
nothing!
What's your position on this, I would be really interested by your views on
the PADS BOF.
As you know, I agree with you that topology awareness is a weak point of
the MIDCOM approach in general. However, there are several existing and
planned networks where awareness is given by other means than by automatic
discovery or where discovery is made outside of the signaling framework.
Take for example the planned 3G mobile access networks that are migrating
to IP. Here I expect that topology awareness will be given, call it remnant
or legacy.
MIDCOM will most likely be shut down (that's also what seems to be Melinda's
What do you mean by shut down? I expect it to complete its charter.
And the MIDCOM protocol itself can be used path-coupled as well as
path-decoupled.
view) and we need to have off path signaling for transitioning to in-path
signaling as initially not all the end host will be supporting in-path
signalling and the evolution could be smoother by having someone to do the
job for them when the topology awareness is not a major pain on the
Middlebox control agent. In addition some network administrators might be
sufficiently happy with PADS and might not even go towards the full blown
architecture leading to path coupled signalling (if their topology is really
simple, most likely this apply to SOHO and residential LANs).
That's my view of PADS and that's what I'll be presenting, in addition my
preference is that PADS BOF serves as input (and not be a separate WG from
NSIS) to the current direction of the NSIS by decoupling the next hop
discovery from the policy rule instalation, this will allow that a subset of
the NSIS protocol could be used for PADS and when the discovery procedure is
used in-path is used.
This will allow implementors of Middlebox control technology on Middleboxes
to do both path coupled and path decoupled in one shot as well.
Obviously there are possible delays to have the decoupling and we would need
to have a back door in case NSIS WG doesn't move fast enough on closing the
decoupling of next hop discovery from policy rule creation on the next hop.
Two points:
1. The NSIS WG should make progress in the base protocol first.
Then the MIDCOM issue can be addressed appropriately. This will certainly
take some more time and it looks like we have to be a bit patient.
2. What do you mean with back door?
Juergen
--
Juergen Quittek quittek@ccrle.nec.de Tel: +49 6221 90511-15
NEC Europe Ltd., Network Laboratories Fax: +49 6221 90511-55
Kurfuersten-Anlage 36, 69115 Heidelberg, Germany http://www.ccrle.nec.de
Sven is organizing a call on Monday from 2 to 3 PM (Sven please correct me
if I'm wrong) and I will be presenting a couple of slides with regards to
PADS usefulness during the transition phase and would be really happy to get
your comments.
Thanks
Cedric
+44 1 628431873
PS: I'm still in France this is my VoIP phone number.
-----Original Message-----
From: sven.van_den_bosch@alcatel.be
[mailto:sven.van_den_bosch@alcatel.be]
Sent: 05 March 2003 16:22
To: ietf@ietf.org; nsis@ietf.org
Cc: brunner@ccrle.nec.de; olov.schelen@operax.com; Aoun, Cedric
[GOLF:MA01:EXCH]; sob@harvard.edu; mankin@psg.com
Subject: path-decoupled signaling (pads) BOF
We have added a mailing list for this BOF and updated the agenda.
Path-decoupled signaling BOF (pads)
Thursday, March 20 at 1530-1730
==============================
CHAIRS: Sven Van den Bosch, sven.van_den_bosch@alcatel.be
Marcus Brunner, brunner@ccrle.nec.de
Mailing list:
pads@psg.com
To subscribe - pads-request@psg.com
Archive - http://psg.com/lists/pads
Description
The Next Steps in Signaling (NSIS) WG is developing a next-generation
general purpose signaling protocol for state installation along the data
path. NSIS is currently focusing on an interpretation of 'along the data
path' as 'visiting the same routers as the data flow'. This is denoted as
path-coupled signaling. For a number of applications, it would be useful to
have a somewhat broader interpretation of 'along the data path', e.g.
'visiting the same AS's as the data flow'. This is denoted as
path-decoupled signaling.
The purpose of the BOF session is to survey and discuss the path-decoupled
signaling case. The first part of the discussion should bring forward
arguments for path-decoupled signaling and describe use cases for
path-decoupled signaling, potentially limiting the scope of the signalling,
e.g. to a single AS.
The second part deals with the expected impact path-decoupled signaling
would have on the NSIS protocol.
Proposed Agenda
Introduction, Chair
Use cases for path-decoupled signaling, Olov Schelen
(olov.schelen@operax.com), Cedric Aoun (cedric.aoun@nortelnetworks.com)
Impact on nsis, TBD
Discussion and round-up
Drafts related to the discussion:
draft-schelen-nsis-opopsig-01
draft-geib-sig-guarandif-00.txt
draft-ietf-nsis-fw-02.txt
draft-declercq-vsn-arch-00.txt