Document: Requirements for Multi-Segment Pseudowire Emulation Edge-to-Edge (PWE3)
Reviewer: Joel Halpern
Review Date: 16 Sept 2007


This document seems to be in a revision-needed state from last June, so it is quite possible that these comments have been overtaken by events.


Summary: This document appears ready for publication as an Informational RFC

Minor comments:
In section 3.2 on architecture the acronym S-PE is used in Figure 2 and the text thereafter with no expansion. It would be a good idea to put in the meaning of S-PE just before it is used, as is done with T-PE. (It is expanded in the next paragraph. The guideline I was taught was to always expand at first use.)

I found the wording of requirement iv (in section 6.1.1) "The MS-PWs MUST be composed of SS-PWs." slightly odd. The rest of the document is written so that one thinks of an MS-PW as composed of segments, and an SS-PW as being composed of a single segment. I think this requirement is intended to say that individual segments behave just like SS-PWs. But I am not sure.

Does the "should" in section 6.3.2 on Statically configured MPLS-PW based MS-PWs (about the need for in-band PW OAM capabilities) belong as a "SHOULD". Note: If this is a "SHOULD" then the best practice is to at least provide an example of when the specified behavior is not required.
The same question apples to "should" in sections 6.3.3 and 6.4.2 on Resiliency.
(These comments on "SHOULD" are prompted by the presence of several capitalized MUST and MAY clauses earlier in the document.)
There appear to actually be a lot of these, including "must" in the rest of section 6. I will not call out each one. It is particularly eye catching when in the list in section 6.4.3 on Quality of service the first two bullet items use "must" and the next two use "MUST"