Document: draft-ietf-pcn-marking-behaviour-03.txt
Reviewer: David L. Black
Review Date: 18 June 2008
IETF LC End Date: 18 June 2008

Summary: 
This draft is basically ready for publication, but has nits that should be fixed before publication.

Comments:

This draft specified requirements on how the diffserv components in a PCN node need to behave in order to support PCN.  In addition to specifying the requirements, an example algorithm (Appendix
A) and extensive implementation notes (Appendix B) are provided.
The implementation notes are very useful - the WG should be commended for working through the practical implementation considerations for this functionality up front as opposed to leaving them to implementers to puzzle out.

All of my comments are on relatively minor points:

"PCN-excess-rate" strikes me as a bad name for a configured throughput rate - I suggest choosing another term.  Intuitively, I would expect PCN-excess-rate to refer to traffic that is in excess of a PCN configured rate.

Section 2.1, 1st paragraph and Section B.3 need to cite a reference for how the DSCP and ECN field values are used to decide whether a packet is PCN or not.

In Section 2.2, first bullet, what is "scheduling rate"?  Please supply a definition.

Section 2.4 - the second bullet also applies to competing non-PCN traffic (if that traffic is metered).  The text needs to be adjusted to allow for this.  Try:

  A packet SHOULD NOT be metered ...

  o If the PCN-packet is already ....

  o If this PCN-node drops the packet.

Section 2.5 could use a reminder that while competing non-PCN packets may be metered, they MUST NOT be marked.  This may have implications for the marking behavior that are probably most usefully discussed in Appendix B.

Nits:
- At the end of the first paragraph in B.1
	"there needs to be" --> "there should be"
  The "should" is lower case.

- idnits 2.11.11 noted that there is now a -04 version of the
	pcn-baseline-encoding draft, as was noted in the draft
	shepherd's writeup.