wiki:WG adoption QA review draft-filsfils-spring-segment-routing

Routing Directorate Quality Assurance (QA) review


Andy Malis, 6 October 2014


In general, the draft is clearly written. The list of companion documents in section 1.1 is appreciated. However, if this section remains through WGLC, the WG should be aware that the RFC Editor will have to process all of the documents that cross-reference each other as a single block. This prevents incremental publication, if that's what the WG intends.

Major Issues

Protocol constructs are used without a reference for where they are defined. For example, the text in sections 3.2 and later discusses the use of an L-Flag, P-Flag, N-Flag, and others, but does not indicate where they are defined. I know that they are defined in draft-ietf-ospf-segment-routing-extensions-00 and the equivalent isis draft, but that is not indicated at appropriate places, such as where the flags are used. Also, draft-ietf-ospf-segment-routing-extensions-00 and the isis draft are listed as informative drafts and they should be normative, as should any other referred draft that defines protocol constructs that are used in a normative manner (not just these two drafts).

Section 3.6.2 discusses two alternative ways to implement tunnel headend processing. For interoperability, one of those should be described as required to implement and the default to use if both are implemented.

There are some TBDs that should have at least an attempt for text before this draft is accepted for WG adoption, especially Security Considerations, or at worst should be addressed early on as a WG draft.

Minor Issues

Number of authors on the first page

RFC 7322 states that the number of authors on the first page should be limited to five individuals. It's difficult to believe that eleven people contributed substantial text to this draft. At the time that this become a WG draft, the list of authors should be reduced to the actively contributing authors (at most five), and the remainder moved to the acknowledgements section.

RFC 2119 Key Words

Section 2, Terminology, contains RFC 2119 key words, which do not belong in a section that just defines the terminology. For example, the active segment definition could just be "the segment that is be used by the receiving router to process the packet."

Definition of a "segment"

Segments are key to the technology, and should receive a better definition than just "a segment identifies an instruction", especially as "instructions" are defined earlier in the document as being "called segments".



A list of abbreviations would be useful for those that are not included in .

"constrained-based path computation" should be "constraint-based path computation".

The document could use a cleanup editing pass, but there was nothing so egregious that it cannot wait until WGLC or even RFC Editor processing.

Last modified 5 years ago Last modified on Oct 6, 2014, 3:02:17 PM