Re: [PCN] New Version Notification fordraft-ietf-pcn-baseline-encoding-01

[<[e-mail]>]

Hi Ruediger,

My argument for making PM an invalid transition for a PCN-ingress node
is that in my head I have a conceptual model that views the PCN region
as consisting of a number of nodes that do PCN marking surrounded by a
fence of PCN-boundary nodes. These boundary nodes simply control ingress
and egress from the region and don't actually do any marking. Of course
in practise the ingress and egress nodes would also have marking
functionality but it is easier to think of them as separate nodes.

As far as I know there is no current proposed extension where EXP can't
be marked to PM. Even in Bob's proposal with fully working tunnels EXP
could be remarked to PM. The question I had in my head was the other way
round: If we don't specify a behaviour for EXP in the baseline will
extension encodings still work if any nodes haven't been upgraded? The
answer to that seemed to be "no" for the majority of proposals (apart
from the simple variant of 3 state encoding).

As you say we also need to decide what to do with EXP packets regarding
metering. I think as currently written Phil's draft allows them to be
metered but I will check that with him.

Toby

-----Original Message-----
From: Geib, Ruediger [mailto:[e-mail]] 
Sent: 21 October 2008 07:47
To: Moncaster,T,Toby,CXR9 R
Cc: [e-mail]
Subject: RE: [PCN] New Version Notification
fordraft-ietf-pcn-baseline-encoding-01

Toby,

thanks for adding more content in the reviewed version. Apart from my 
support of Michaels proposal 1., I still would like to comment on 
some of your new text: 

Appendix B, Table 2, row "ingress"

If "PM" is an invalid codepoint out, then the ingress node does not 
belong to the PCN domain. As far as I understood, indication of 
pre-congestion on a link is signaled by a PM mark. The ingress 
node following your specification in row "ingress" however 
can't signal pre-congestion.

Specifying behaviours for unexpected codepoints is a necessity and 
writing down this requirement adds value to the document. The 
question is, whether forward compatibility benefits from handling 
EXP as NM. The whole topic is in fact a marking issue, and there
it should be dealt with. Thoughts on your proposal:
- Could a future extension exist, where there's no transition 
  from EXP to PM? Then your proposal is harmful.
- What you don't discuss, as you are writing the encoding draft,
  is whether the EXP packets should be metered by an interior 
  node just supporting baseline encoding. I think they should be. 
  But may be, I'm wrong and all of this discussion on receiving 
  packets with unexpected codings and backward/forward 
  compatibility should be a requirement for a PCN deployment 
  scenario document.

Regards,

Ruediger


-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of
[e-mail]
Sent: Tuesday, October 14, 2008 4:52 PM
To: [e-mail]
Subject: Re: [PCN] New Version Notification
fordraft-ietf-pcn-baseline-encoding-01

As promised I have uploaded a new version. I am aware there are a
handful of typos and spelling mistakes that I noticed just after
uploading. I will correct these in the next version. I am now pretty
happy with almost all the text.

The key section for the WG to note is Appendix B where I have put in a
couple of questions that I would welcome input on. 

The first of these is where to put the text (now set out as a table)
about valid and invalid transitions at PCN nodes? Should this be part of
the baseline encoding or should it be split between a new edge node
behaviour document and Phil's marking behaviour.

The second question relates to how to best handle the unexpected
presence of the EXP codepoint. There is a potential for this to happen
during any partial deployment of an experimental encoding extension
scheme. Whilst it would be best for an operator to ensure all nodes run
the same encoding there is a chance this might not happen. The best
solution as I see it is for the baseline scheme to treat EXP the same as
NM but to raise a management alarm...

I have also highlighted another point in Appendix A which needs a
decision: should we as a WG recommend that RFC3168 full functionality
tunnels SHOULD NOT be used in a PCN-domain or would it be better to put
our collective voices behind Bob's attempts to correct the anomalous
behaviour of tunnels which he is currently trying to push through TSVWG
(draft-briscoe-tsvwg-ecn-tunnel-01)

Toby

-----Original Message-----
From: IETF I-D Submission Tool [mailto:[e-mail]] 
Sent: 14 October 2008 15:35
To: Moncaster,T,Toby,CXR9 R
Cc: Briscoe,RJ,Bob,CXR9 R; [e-mail]
Subject: New Version Notification for
draft-ietf-pcn-baseline-encoding-01 


A new version of I-D, draft-ietf-pcn-baseline-encoding-01.txt has been
successfuly submitted by T Moncaster and posted to the IETF repository.

Filename:	 draft-ietf-pcn-baseline-encoding
Revision:	 01
Title:		 Baseline Encoding and Transport of Pre-Congestion
Information
Creation_date:	 2008-10-14
WG ID:		 pcn
Number_of_pages: 12

Abstract:
Pre-congestion notification (PCN) provides information to support
admission control and flow termination in order to protect the
Quality of Service of inelastic flows.  It does this by marking
packets when traffic load on a link is approaching or has exceeded a
threshold below the physical link rate.  This document specifies how
such marks are to be encoded into the IP header.  The baseline
encoding described here provides for only two PCN encoding states.
It is designed to be easily extended to provide more encoding states
but such schemes will be described in other documents.
 



The IETF Secretariat.


_______________________________________________
PCN mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/pcn
_______________________________________________
PCN mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/pcn