RE: [PCN] Re: ECN support in a PCN domain

<philip.eardley@bt.com> Wed, 28 November 2007 17:49 UTC

Return-path: <pcn-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1IxR2g-0002KC-84; Wed, 28 Nov 2007 12:49:46 -0500
Received: from pcn by megatron.ietf.org with local (Exim 4.43) id 1IxR2f-0002GP-8I for pcn-confirm+ok@megatron.ietf.org; Wed, 28 Nov 2007 12:49:45 -0500
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1IxR2e-0002GF-UX for pcn@ietf.org; Wed, 28 Nov 2007 12:49:44 -0500
Received: from smtp2.smtp.bt.com ([217.32.164.150]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IxR2d-0003GU-3F for pcn@ietf.org; Wed, 28 Nov 2007 12:49:44 -0500
Received: from E03MVZ1-UKDY.domain1.systemhost.net ([193.113.30.65]) by smtp2.smtp.bt.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 28 Nov 2007 17:49:41 +0000
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [PCN] Re: ECN support in a PCN domain
Date: Wed, 28 Nov 2007 17:49:41 -0000
Message-ID: <75A199C5D243C741BF3D3F1EBCEF9BA503B343D6@E03MVZ1-UKDY.domain1.systemhost.net>
In-Reply-To: <474D4683.4080209@ericsson.com>
X-MS-Has-Attach:
X-MS-TNEF-Correlator:
Thread-Topic: [PCN] Re: ECN support in a PCN domain
Thread-Index: Acgxq7MTUQ7YyIB6S/Sr7L4i/hUjJQANz74A
From: philip.eardley@bt.com
To: magnus.westerlund@ericsson.com, steven.blake@ericsson.com
X-OriginalArrivalTime: 28 Nov 2007 17:49:41.0935 (UTC) FILETIME=[0DD553F0:01C831E7]
X-Spam-Score: -1.0 (-)
X-Scan-Signature: b5d20af10c334b36874c0264b10f59f1
Cc: pcn@ietf.org
X-BeenThere: pcn@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: PCN WG list <pcn.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/pcn>, <mailto:pcn-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/pcn>
List-Post: <mailto:pcn@ietf.org>
List-Help: <mailto:pcn-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/pcn>, <mailto:pcn-request@ietf.org?subject=subscribe>
Errors-To: pcn-bounces@ietf.org

> >
> >>>    o  If a packet that is part of a PCN-flow arrives at a
PCN-ingress-
> >>>       node with its CE (Congestion experienced) codepoint set,
then we
> >>>       assume that the PCN-ingress-node drops the packet.  After
its
> >>>       initial Charter is complete, the WG may decide to work on a
> >>>       mechanism (such as through a signalling extension) that
enables
> >>>       ECN-marking to be carried transparently across the
PCN-domain."
> >> This one really surprise me. How can you make this assumption? In
case
> a
> >> ECN enabled flow with some packets CE marked arrive to the ingress
of
> >> the PCN domain and are within the limits for which the flow has
been
> >> admitted into the PCN domain, then the PCN domain has no right to
> >> summarily drop that packet as being less important than any of the
> >> unmarked packets. Only in cases if congestion actually is
experience
> >> within the PCN domain would a differential treatment of CE marked
> >> packets being warranted.
> >>
> >> Can someone please explain how the above assumptions still satisfy
the
> >> following sentence in the charter?
> >>
> >> "All PCN mechanisms, including transport and encoding
> >> of (pre-congestion) information, are required to cleanly integrate
> >> with existing architectures and protocols such as DiffServ and
ECN."
> >
> > I think that PCN has been defined in the charter to not apply for
> > traffic which might use an ECN-capable transport protocol.
> >
> 
> No, I don't agree with this characterization. I can't remember that
> being the notion. If I remember Lars and my discussion one of the
> reason we did charter PCN like we did was to ensure that you actually
> could find a encoding that works and maintains ECN across the PCN
> domain. And that is according to my memory what the following sentence
> is intended for:
> 
> All PCN mechanisms, including transport and encoding
> of (pre-congestion) information, are required to cleanly integrate
> with existing architectures and protocols such as DiffServ and ECN.
> 
> And if there exist no reasonable solution to encode PCN while
> maintaining ECN across a PCN domain. Well that would be tough luck and
> we would have to revisit the fundamentals. Which could require a new
BOF
> and rechartering to remove those.
> 
> My main message is: As I see it you are not allowed to take the easy
way
> out and ignore that incomming traffic may be ECN enabled. You will
have
> to do something reasonable with that.
> 
magnus,
here's the background to the bulleted text [at the start of above - in
current I-D]. or rather, my perspective on the background.

there was quite a lot of discussion (without resolution) whether it's
reasonable for a flow that uses PCN adm ctrl to also want e2e ECN for
rate adaptation (discussion both in terms of whether a flow would
actually ever want this, and how well the PCN mechanism would work if
flows were adapting like this). 
this was resolved by saying that the support of e2e ecn is out of scope
of initial charter of pcn. 
this leaves the question: if/when after re-chartering the scope does
include e2e ecn, how would this impact pcn encoding choice (which we
have to make now)? 
The solution: 
- we know that if use (only) DSCPs for encoding PCN-marks, then e2e ECN
would be ok
- we know that if use ECN field for encoding PCN-marks, then e2e ECN
could be supported; the obvious way is to tunnel the flow over the
PCN-domain & to establish the tunnel as part of the signalling set-up
for this flow. 
- [rationale for the text at top of this email quoted from the I-D]: but
we haven't worked out the details for the tunnelling /signalling, so
let's give some very simple advice if the situation does occur. The best
thing seems to drop the pkt, as at least this means the receiver behaves
in the same way as if it got the ecn-marked pkt. If this is felt to be
unhelpful or misleading, we could instead say something about
signalling/tunnelling approach ['and details would be worked out after
re-chartering as appropriate..']

phil


_______________________________________________
PCN mailing list
PCN@ietf.org
https://www1.ietf.org/mailman/listinfo/pcn