[Pals] WGLC comments on draft-ietf-pals-rfc4447bis

[Stewart Bryant <[e-mail]>]

Here are my WGLC comments on draft-ietf-pals-rfc4447bis-01.txt


SB> You should note in the metadata, abstract and intro that this
SB> obsoletes RFC4777

  Abstract


SB> You have a CW change that should be noted here.


1. Introduction

SB> This is the original intro. A number of things have moved
SB> on and people are more familiar with the technology
SB> than when you first wrote this.
SB> It is worth reflecting on whether this is still the
SB> right introduction to use 10 years after publication
SB> and 13 years since the first WG draft.


    QoS-related issues are not discussed in this document.

SB> QOS needs to be expanded.


    For the purpose of this document, PE1 will be defined as the ingress
    router, and PE2 as the egress router.  A layer 2 PDU will be received
    at PE1, encapsulated at PE1, transported and decapsulated at PE2, and
    transmitted out of PE2.

SB> Of course it's not just L2 packets - since you have IP PWs
SB> and I am not sure that TDM is really L2.

  Note that this document was written to address errata in [RFC4447].



4. Details Specific to Particular Emulated Services

4.1. IP Layer 2 Transport

    This mode carries IP packets over a pseudowire.  The encapsulation
    used is according to [RFC3032].  The PW control word MAY be inserted
    between the MPLS label stack and the IP payload.  The encapsulation
    of the IP packets for forwarding on the attachment circuit is
    implementation specific, is part of the native service processing
    (NSP) function [RFC3985], and is outside the scope of this document.

SB> You might consider just having section 4, rather than empty
SB> 4 followed by small section 4.1



6.2. PW Types for which the Control Word is NOT mandatory

SB> NOT mandatory is not correct RFC2119 - I think you mean NOT REQUIRED

    If a system is capable of sending and receiving the control word on
    PW types for which the control word is not mandatory, then each such
    PW endpoint MUST be configurable with a parameter that specifies
    whether the use of the control word is PREFERRED or NOT PREFERRED.

SB> (NOT) PREFERRED are not RFC2119 complaint, we need to either use
SB> RFC2119 terms of remove the capitals or put them in quotes
SB> if the capitalization is formal notation.

    For each PW, there MUST be a default value of this parameter.  This
    specification does NOT state what the default value should be.

SB> I don't think NOT should be capitalised (same below)

    If a system is NOT capable of sending and receiving the control word
    on PW types for which the control word is not mandatory, then it
    behaves exactly as if it were configured for the use of the control
    word to be NOT PREFERRED.

SB> 2119 again with NOT PREFERRED








9. Changes from RFC4447

    The changes in this document are mostly minor fixes to spelling and
    grammar, or clarifications to the text, which were either noted as
    errata to RFC4447 or found by the editors.

    However a new section (6.3) on control-word renegotiation by label
    request message has been added, referencing RFC 6723.   The diagram
    of C-bit handling procedures has also been removed, as the updated
    diagram in RFC 6723 is now definitive.

SB> This should be reflected in the Abstract, and probably moved up to
SB> the introduction, or at least a pointer placed in the introduction.





14. Additional Historical Contributing Authors


SB> It is convention to put additional names before the authors/editors
SB> so that that the authors are at the end of the document.
SB>
SB> In many cases we know that the original authors have changed
SB> their details, and I wonder whether it would not be appropriate
SB> to at least say "now at foo corp, email address bar"

- Stewart