RE: Gen-ART Last Call Review of draft-ietf-avt-hc-over-mpls-protocol-07

["ASH, GERALD R \(JERRY\), ATTLABS" <[e-mail]>]

Hi Spencer,

Many thanks for your thorough and constructive review of the draft.  

Please see responses to your comments below, and please let us know of
any further comments or suggestions.

Thanks,
Regards,
Jerry 

> -----Original Message-----
> From: Spencer Dawkins [mailto:[e-mail]] 
> Sent: Wednesday, February 07, 2007 12:20 PM
> To: [e-mail]; [e-mail]; 
> [e-mail]; HAND, JAMES, ATTLABS; ASH, GERALD R 
> (JERRY), ATTLABS; Mark Townsley; ext Cullen Jennings
> Cc: [e-mail]; General Area Review Team
> Subject: Gen-ART Last Call Review of 
> draft-ietf-avt-hc-over-mpls-protocol-07
> 
> I have been selected as the General Area Review Team (Gen-ART)
> reviewer for this draft (for background on Gen-ART, please see
> http://www.alvestrand.no/ietf/gen/art/gen-art-FAQ.html).
> 
> Please resolve these comments along with any other Last Call comments
> you may receive.
> 
> Document: draft-ietf-avt-hc-over-mpls-protocol-07
> Reviewer: Spencer Dawkins
> Review Date: 2007-01-30
> IETF LC End Date: 2007-07-07
> IESG Telechat date: (not known)
> 
> Summary: This document is on the right track for publication 
> as Proposed 
> Standard. I had some questions (please see below), but the 
> quality seemed 
> very good (thanks for that).
> 
> I'd like to see some work on my comments in 5.1, 5.2 
> (especially 5.2.1), and 
> 5.3. I had some comments on clarity in section 6, but these were 
> more-than-nits-but-not-problems.
> 
> Thanks!
> 
> Spencer
> 
> Comments:
> 
> 1. Introduction
> 
>    Voice over IP (VoIP) typically uses the encapsulation
>    voice/RTP/UDP/IP.  When MPLS labels [RFC3031] are added, 
>    this becomes
>    voice/RTP/UDP/IP/MPLS-labels.  MPLS VPNs (e.g., [RFC2547]) 
>    use label
>    stacking, and in the simplest case of IPv4 the total 
>    packet header is
>    at least 48 bytes, while the voice payload is often no more than 30
>    bytes, for example.  When IPv6 is used, the relative size of the
>    header in comparison to the payload is even greater.  The 
>    interest in
>    header compression (HC) is to exploit the possibility of
>    significantly reducing the overhead through various compression
>    mechanisms, such as with enhanced compressed RTP (ECRTP) [RFC3545]
>    and robust header compression (ROHC) [RFC3095], and also 
>    to increase
>    scalability of HC.  MPLS is used to route HC packets over an MPLS
>    label switched path (LSP) without compression/decompression cycles
>    at each router.  Such an HC over MPLS capability can increase
>    bandwidth efficiency as well as the processing scalability of the
>    maximum number of simultaneous compressed flows that use HC at each
>    router.  Goals and requirements for HC over MPLS are discussed in
>    [RFC4247].  The solution put forth in this document using MPLS
>    pseudowire (PW) technology has been designed to address these goals
>    and requirements.
> 
> Spencer (Nit): I think the last sentence is actually "The 
> solution using MPLS pseudowire
> (PW) technology put forth in this document has been designed 
> to address
> these goals and requirements." (the solution wasn't actually 
> put forth using MPLS PW technology :-)

Agree with your suggested rewording.
 
> 2. Contributors
> 
>    Besides the editors listed in Section 12, the following people
>    contributed to the document:
> 
> Spencer (Nit): I like the use of this section, but it seems 
> odd to have it
> so far from the acknowledgements section. I'm not sure if 
> IESG has an agreed
> sense of taste for placement or not. Cullen?

In a recent RFC review I saw the RFC editor put the "Contributing
Authors" section right before the "Editor's Address" section, where both
sections were toward the end of the RFC (near the "Acknowledgements"
section).  It seems like a good approach IMO.
 
> 3. Terminology
> 
>    PSN Tunnel Signaling: Used to set up, maintain, and tear down the
>    underlying PSN tunnel
> 
> Spencer (Nit): s/Used/A protocol used/ (all of the other 
> definitions look
> like complete sentences, this one is a fragment)

OK.
 
> 5.1 MPLS Pseudowire Setup & Signaling
> 
>    This specification defines new PW type values to be carried within
>    the FEC object to identify HC PWs for each HC scheme.  The 
>    PW type is
>    a 15-bit parameter assigned by IANA, as specified in the [RFC4446]
>    registry, and MUST be used to indicate the HC scheme being used on
>    the PW.  The following PW type values have been set aside for
>    assignment by IANA:
> 
>    0x001A  ROHC Transport Header-compressed Packets    [RFC3095]
>    0x001B  ECRTP Transport Header-compressed Packets   [RFC3545]
>    0x001C  IPHC Transport Header-compressed Packets    [RFC2507]
>    0x001D  cRTP Transport Header-compressed Packets    [RFC2508]
> 
> Spencer: "have been set aside for assignment by IANA", with 
> RFC references,
> confused me badly here. I read this text as saying that these 
> values were
> set aside IN [RFC3095], etc, which was wrong. Perhaps "IANA 
> is requested to assign the following new PW type values:"?

The language is based on exchanges with Luca Martini icw RFC 4446.  The
registry is specified in RFC 4446
http://www.ietf.org/rfc/rfc4446.txt?number=4446.  Using the text you
suggested in the email exchange with Andy:

"IANA has set aside the following PW type values for assignment
according to the registry specified in RFC 4446, Section 3.2:

  PW type Description                                 Reference
  =============================================================
  0x001A  ROHC Transport Header-compressed Packets    [RFC3095]
  0x001B  ECRTP Transport Header-compressed Packets   [RFC3545]
  0x001C  IPHC Transport Header-compressed Packets    [RFC2507]
  0x001D  cRTP Transport Header-compressed Packets    [RFC2508]"

>    The PW control word enables distinguishing between various packets
>    types (e.g., uncompressed, UDP compressed, RTP compressed,
>    context-state, etc.).  However, the PW control word indications are
>    not unique across HC schemes, and therefore the PW type 
>    value allows
>    the HC scheme to be identified.
> 
> 5.2 Header Compression Scheme Setup, Negotiation, & Signaling
> 
>    Pseudowire Interface Parameter Sub-TLV type values are specified in
>    [RFC4446].  Two code-points have been reserved, as follows:
> 
>    Parameter ID Length        Description
References
>    0x0D      up to 256 bytes  ROHC over MPLS configuration    RFC 3241
>    0x0F      up to 256 bytes  CRTP/ECRTP/IPHC HC over MPLS    RFC 3544
>                               configuration
> 
> Spencer: Again, something like "IANA is requested to assign these new
> code-points" would help me with what seems to be ambiguous text.

Again, suggest this clarifying text:

"IANA has set aside the following Pseudowire Interface Parameter Sub-TLV
type values according to the registry specified in RFC 4446, Section
3.3:

  Parameter ID Length        Description                     References
  0x0D      up to 256 bytes  ROHC over MPLS configuration    RFC 3241
  0x0F      up to 256 bytes  CRTP/ECRTP/IPHC HC over MPLS    RFC 3544
                             configuration"

> 5.2.1 Configuration Option Format [RFC3544]
> 
>    Both the network control protocol for IPv4, IPCP [RFC1332] and the
>    IPv6 NCP, IPV6CP [RFC2472] may be used to negotiate IP HC 
>    parameters
>    for their respective protocols.  The format of the configuration
> 
> Spencer (Nit): I understand what you're saying with "their respective
> protocols", but I don't think what you're saying, and what 
> I'm hearing, is
> actually what these words mean. Perhaps "their respective controlled
> protocols"? That would be unambiguous.

Agree with your proposed rewording.
 
>    option is the same for both IPCP and IPV6CP.  This configuration
>    option MUST be included for ECRTP, CRTP and IPHC PW types and MUST
>    NOT be included for ROHC PW types.
> 
> Spencer: Is it obvious what the decompressor does if it sees this
> configuration option for ROHC PW types? It may be - I'm just 
> asking. I'd
> have the same question elsewhere (in 5.2.2, for example), but 
> will only ask it here.

Yes.  The corresponding text for the ROHC configuration option is
specified in Section 5.2.5.  In other sections we specify that the
configuration options are only applicable to specific header compression
formats, e.g., as in Section 5.2.2 for cRTP.
 
> 5.2.3 Enhanced RTP-Compression Suboption [RFC3544]
> 
>    To use the enhanced RTP HC defined in [RFC3545], a
>    new sub-option 2 is added.  Sub-option 2 is negotiated instead of,
>    not in addition to, sub-option 1.  This suboption MUST be included
> 
> Spencer: is it obvious what the decompressor does if it sees 
> a compressor specifying sub-option 1 AND sub-option 2?

"Sub-option 1" refers to the RTP-Compression Suboption, as specified in
Section 5.2.2, and "sub-option 2" refers to the Enhanced RTP-Compression
Suboption, as specified in Section 5.2.3.  These suboptions do not occur
together.  I agree that this should be clarified in the text.
 
>    for ECRTP PWs (0x001B) and MUST NOT be included for other PW types.
> 
>    MAX_HEADER
> 
>       NOTE: The four ROHC profiles defined in RFC 3095 do not provide
>       for a MAX_HEADER parameter.  The parameter MAX_HEADER defined by
>       this document is therefore without consequence in these 
>       profiles.  Other profiles (e.g., ones based on RFC 2507) can
make 
>       use of the parameter by explicitly referencing it.
> 
> Spencer: is there any statement you can make about the 
> largest header that
> can be compressed in these profiles? Even if not, it might be 
> good to say so
> explicitly - I don't know what "without consequence" says about what
> implementations may encounter in practice.

Have to look into whether there is largest header that can be compressed
in these ROHC profiles.  I agree that the 'without consequence' phrasing
should be clarified.
 
> 5.4 Packet Reordering
> 
>    Packet reordering for ROHC is discussed in [RFC4224], which is a
>    useful source of information.  In case of lossy links and other
>    reasons for reordering, implementation adaptations are needed to
>    allow all the schemes to be used in this case.  Although CRTP is
>    viewed as having risks for a number PW environments due to 
>    reordering
> 
> Spencer (Nit): "a number of PW environments" ("of" is missing)

OK.
 
>    and loss, it is still the protocol of choice in many 
>    cases.  In these
>    circumstances, it must be implemented and deployed with care.  IPHC
>    should use TCP_NODELTA, ECRTP should send absolute values, ROHC
>    should be adapted as discussed in [RFC4224].  An evaluation and
>    simulation of ECRTP and ROHC reordering is given in [REORDER-EVAL].
> 
> Spencer (Probably a Nit): It wasn't obvious to me whether these
> recommendations are sufficient to "implement and deploy with care", or
> whether additional precautions must be taken. Even putting these
> recommendations in a numbered list immediately after 
> "deployed with care"
> would be sufficient, if these recommendations are sufficient.

This goes back to a discussion with Allison Mankin RE CRTP issues
discussed icw RFC 4446.  There was no further list of recommendations
out of that discussion, rather, the point is that in packet-lossy
environments, for example, CRTP may not work well and ECRTP may perform
better.  Some folks felt that CRTP should be excluded because of that
problem.  There were, however, other concerns raised on deploying ECRTP
(e.g., CRTP is already widely deployed, plus other reasons).
 
> 6. HC Pseudowire Setup Example
> 
>    The Label Mapping message sent from R4/HD to R1/HC would be almost
>    identical to the one sent in the opposite direction, with the
>    following exceptions
> 
> Spencer (Nit): missing a colon here...

OK.

>    As soon as either R1/HC or R4/HD had both transmitted and received
>    Label Mapping Messages with the same PW Type and PW ID, it could
> 
> Spencer: "it" is not entirely clear here. "It" usually refers 
> to one thing 
> (in my mind), and the subject of the sentence is two things 
> ("either R1/HC 
> or R4/HD"). Mumble. Would it help to say that each ROHC 
> endpoint considers 
> the PW established when it has seen both packets?

Agree with your suggested rewording.
 
>    consider the PW established.  R1/HC could send ECRTP packets using
>    the label it received in the Label Mapping Message from R4/HD, Lr4,
>    and could identify received ECRTP packets by the label it 
>    had sent to
>    R4/HD, Lr1.  And vice versa.
> 
>    The following 3 RTP packets from this flow would be sent as
> 
> Spencer: "following" is not entirely clear here. Suggest "next"?

OK.
 
>    COMPRESSED_UDP_8, to establish the absolute and delta values of the
>    IPv4 identifier and RTP timestamp fields.  These packets would use
>    the same ECRTP CID as the previous 3 FULL_HEADER packets.  The MPLS
>    and PW headers at the beginning of these packets would be formatted
>    as follows:

--- Begin Message ---

Hi,

In a bit of off-list discussion, we had two threads:

1. would it be practical to assume that the RFC 3544
   protocol would be present for IPHC/CRTP eCRTP?  The answer
   is that the implementations are not restricting themselves
   to that design, so Kristofer Sandlun's thoughtful codepoint 
   suggestion isn't an option.

2. are there some very pressing reasons why vendors cannot
   use eCRTP and find themselves using CRTP?  The answer to
   this turns out to be yes, and turns out to be
   sometimes IPR, which is certainly not an unknown finding
   for IETF...

I agreed to put cRTP back in the registrations with a note
that reflects that the use of cRTP has to be an informed tradeoff
(because the later IETF documents say that cRTP has risks for
general use).  I've cc'd the AVT chairs because they may want
to take the new protocol takeup info into account.  Of course,
no specifics - that's also not unknown for IETF.

Here's Yet Another List of these - maybe we are looking at our last one
:)

------
   0x001A  ROHC Transport Header-compressed Packets         [RFC3095, 
 
draft-ietf-rohc-over-reordering-03.txt]
   0x001B  eCRTP Transport Header-compressed Packets        [RFC3545]
   0x001C  IPHC Transport Header-compressed Packets         [RFC2507]
   0x001D  cRTP Transport Header-compressed Packets
[RFC2508][Note 1]



[Note 1]   Although CRTP is viewed as having risks for a number
  PW environments due to misordering and loss, it is registered
  because of markets where commercial issues lead to its choice
  In these circumstances, it must be implemented and deployed
  with care.

-- Add Informational Reference to draft-ietf-rohc-over-reordering-03.txt
------

I added the i-d that gives the advice about improving ROHC for
reordering to the references for ROHC.  This was a good idea derived 
from Kristopher's email.

I couldn't decide there was urgency to record Kristofer's additional
guidance
IPHC and eCRTP here.  But his expertise on these convinced me that IPHC
can
be there without causing trouble, 

So I hope this is it for this material.

Now, about part 1 of my Discuss, if Luca or the group will respond, I'd
love to settle this.

Allison


_______________________________________________
pwe3 mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/pwe3

--- End Message ---

_______________________________________________
Ietf mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/ietf